• Shortcuts : 'n' next unread feed - 'p' previous unread feed • Styles : 1 2

» Publishers, Monetize your RSS feeds with FeedShow:  More infos  (Show/Hide Ads)


Date: Sunday, 21 Sep 2014 14:19

For academic researchers who are readers of this blog (and I know you’re lurking out there), I wanted to call attention to my new paper on Macroeconomic Regimes and Regime Shifts:

Many economic time series exhibit dramatic breaks associated with events such as economic recessions, financial panics, and currency crises. Such changes in regime may arise from tipping points or other nonlinear dynamics and are core to some of the most important questions in macroeconomics. This chapter surveys the literature on regime changes. Section 1 begins with an interpretation of the move of an economy into and out of recession as an example of a change in regime and introduces some of the basic tools for analyzing such phenomena. Section 2 provides a detailed overview of econometric methods that are appropriate for time series that are subject to changes in regime. Section 3 summarizes the ways in which changes in regime can be incorporated into theoretical economic models and briefly reviews applications in a number of areas of macroeconomics.

Author: "James_Hamilton" Tags: "financial markets, here and there, reces..."
Send by mail Print  Save  Delicious 
Date: Friday, 19 Sep 2014 22:46

The MacIver Institute, an organization of endlessly imaginative analysis, has highlighted this LFB memo that reports that under the right conditions, the structural budget balance will be +$535 for the 2015-17 biennium.

Those assumptions include a $116 million cut to this FY’s appropriations, revenues in the 2015-17 biennium would rise (annually) at the rate it has during the previous five fiscal years, 2.9%, and net appropriations in each of the years during the biennium would stay at FY2014-15 levels, “adjusted for one time commitments and 2015-17 commitments.”

I am not an expert in the intricacies of budgeting (here’s a start), and the evolution of Wisconsin tax revenues. However, what I can see in graph for Wisconsin here, for the period 2008-12, suggests to me that 2.9% figure is highly sensitive to sample period (gee, wonder why they picked that particular five year period?). If I use the data at the Governing website, I get a little less than 0.5% growth per annum.

In addition, the zero spending growth assumption is highly unrealistic. As Jon Peacock at the Wisconsin Budget Project wrote:

Some people who derided structural deficits in the past are now arguing that this isn’t a big deal because the state can grow its way out of this problem. That’s true in a sense, but also very misleading. Assuming tax collections increase as expected to about $14.4 billion in the current fiscal year, growth of 4% per year in 2015-17 would close the budget hole if total spending is frozen. But keep in mind that the spending needed for a status quo or “cost to continue” budget typically increases almost as fast as revenue – because of inflation and population growth. Thus, freezing spending in 2015-17 at the current level would not be a painless exercise; it would require significant cuts in areas like Medicaid, K-12 and higher education, and the corrections system budget.

It’s also important to keep in mind that the current structural deficit calculations focus only on the General Fund and assume that in 2015-17 the state will stop transferring dollars from the General Fund to the Transportation Fund. In light of the problems in state and federal financing for transportation, there will be significant pressure to continue to make those transfers.

In other words, the LFB tabulated at the direction of State Representative John Nygren what would happen if one let revenues move, but not spending. Mechanically, it must be that the balance looks better — no mystery there. It’s a well known trick, used earlier on a national stage; for more on the national version of the can opener assumption, see these posts on Ryan plan (I) and Ryan plan (II). For more on MacIver Institute analyses, see this post.

So, for me, a more honest appraisal of the situation is presented in Figure 1 below.

wibalance4

Figure 1: (Negative of) General Fund Amounts Necessary to Balance Budget, by Fiscal Year, in millions of dollars (blue bars); and estimate taking into account shortfall of $281 million for FY2013-14 (red square), and adding $380 million to each of the fiscal years in the 2015-17 biennium (green squares). “Structural” denotes ongoing budget balance, assuming no revenue/outlay change associated with economic growth. Source: Legislative Fiscal Bureau (September 8, 2014), Wisconsin Budget Project, “Wisconsin needs $760 million more for Medicaid,” Channel 3000 and author’s calculations.

Author: "Menzie Chinn" Tags: "Wisconsin"
Send by mail Print  Save  Delicious 
Date: Thursday, 18 Sep 2014 16:40

DWD released August employment figures today. Attainment of Governor Walker’s 250,000 net new private sector jobs continues to be unlikely.

aug14pix0

Figure 1: Private nonfarm payroll employment, s.a., in ‘000s (red), forecast from March 2014 Wisconsin Economic Outlook, interpolated from quarterly, using quadratic match (green), and linear trend consistent with Governor Walker’s pledge to create 250,000 net new private sector jobs by end of first term (gray). Source: Department of Workforce Development, Wisconsin Economic Outlook (March 2014), and author’s calculations.

Total employment did increase, attributable to a large increase in local government employment. Since August is related to the beginning of the school year, the government employment preliminary number is not particularly informative, given the difficulty in seasonal adjustment.

DWD also released Quarterly Census of Employment and Wages data through March; these figures for private employment were no different from figures released earlier, and discussed in this post. At that time, I characterized the outcome thus: “No succor from the QCEW series that the Walker Administration previously touted”.

According to BLS data, based on annual changes, Wisconsin ranked 33rd in terms of private job creation. [2]

Author: "Menzie Chinn" Tags: "Wisconsin"
Send by mail Print  Save  Delicious 
Date: Thursday, 18 Sep 2014 04:35

Wisconsin Department of Health Services indicates additional $760 million Medicaid expenditures for FY15-17. [1] This implies a further worsening of the 2015-17 biennium structural budget balance from -$1.8 billion (discussed here) to something like -$2.56 billion.

wibalance4

Figure 1: (Negative of) General Fund Amounts Necessary to Balance Budget, by Fiscal Year, in millions of dollars (blue bars); and estimate taking into account shortfall of $281 million for FY2013-14 (red square), and adding $380 million to each of the fiscal years in the 2015-17 biennium (green squares). “Structural” denotes ongoing budget balance, assuming no revenue/outlay change associated with economic growth. Source: Legislative Fiscal Bureau (September 8, 2014), Wisconsin Budget Project, “Wisconsin needs $760 million more for Medicaid,” Channel 3000 and author’s calculations.

In addition to a deteriorating budget situation, employment and output are lagging [National Journal}. BLS will report Wisconsin August employment on Friday, DWD I am guessing will release figures tomorrow. I do not expect anything to change my view that by January 2015, employment will undershoot Governor Walker’s promise of 250,000 net new private sector jobs by about 120,000.

Author: "Menzie Chinn" Tags: "Wisconsin"
Send by mail Print  Save  Delicious 
Date: Tuesday, 16 Sep 2014 18:56

That’s a title from an oped by former GW Bush speechwriter and current AEI scholar Marc Thiessen nearly a year ago. We can now evaluate whether in fact the implementation of individual insurance mandate component of the ACA did implode. From “New Data Show Early Progress in Expanding Coverage, with More Gains to Come,” White House blog today:

Figure1_healthinsurance

Source: “New Data Show Early Progress in Expanding Coverage, with More Gains to Come,” White House blog (9/16/2014).

From the post:

…today’s results from the NCHS’ National Health Interview Survey (NHIS) show that the share of Americans without health insurance averaged 13.1 percent over the first quarter of 2014, down from an average of 14.4 percent during 2013, a reduction corresponding to approximately 4 million people. The 13.1 percent uninsurance rate recorded for the first quarter of 2014 is lower than any annual uninsurance rate recorded by the NHIS since it began using its current design in 1997.

As striking as this reduction is, it dramatically understates the actual gains in insurance coverage so far in 2014. The interviews reflected in today’s results were spread evenly over January, February, and March 2014. As a result, the vast majority of the survey interviews occurred before the surge in Marketplace plan selections that occurred in March; 3.8 million people selected a Marketplace plan after March 1,with many in the last week before the end of open enrollment on March 31. Similarly, these results only partially capture the steady increase in Medicaid enrollment during the first quarter.

For this reason, private surveys from the Urban Institute, Gallup/Healthways, and the Commonwealth Fund in which interviews occurred entirely after the end of open enrollment have consistently shown much larger gains in insurance coverage. An analysis published last month in the New England Journal of Medicine based on the Gallup/Healthways data estimated that coverage gains reached 10.3 million by the middle of 2014.

Some points that will surely annoy some commentators, particularly those who bewailed the early low rates of enrollment of young adults:


  • Increases in coverage were concentrated among non-elderly adults, with the youngest adults seeing the largest gains: Insurance coverage increased among non-elderly adults increased by 2.0 percentage points from 2013 through the first quarter of 2014, while insurance coverage was essentially unchanged for children and individuals over the age of 65. This pattern reflects the fact that children and older adults had much greater access to insurance coverage prior to reform.Young adults ages 18-24 saw their uninsurance rate drop by 5.5 percentage points from 2013 to the first quarter of 2014, a much larger decline than seen by other groups.

    The gains in this report are on top of very large coverage gains among young adults from 2010 through 2013, gains that are largely attributable to a provision of the Affordable Care Act that permits young adults to remain on their parents’ insurance policies until they turn 26. From 2010 through the first quarter of 2014, the insurance rate among young adults ages 19-25 (the full group affected by this earlier Affordable Care Act provision) has increased by a total of 13 percentage points.

  • Blacks and Latinos saw particularly large increases in coverage: Insurance coverage among non-Hispanic blacks increased by 3.8 percentage points from 2013 to the first quarter of 2014, while coverage among Latinos increased by 3.1 percentage points. Both increases were considerably larger than 1.3 percentage point increase in the full population.

Had states like Wisconsin expanded Medicare coverage, the uninsured rate would have declined further.

Returning to Marc Thiessen’s prognostications:

President Obama may have no choice but to delay the individual mandate. As my American Enterprise Institute colleague, Dr. Scott Gottlieb, points out, how can Obama penalize people for not having health insurance if the government’s Web site to provide that insurance doesn’t work?

Without the individual mandate, Obamacare unravels. The only way the law works is if the government forces young, healthy people into it by threatening them with penalties for not carrying health insurance. But if there is no penalty for not signing up, then fewer Americans will sign up.

Even if the administration manages to fix the Web site and finally implement the individual mandate, people still may not join — because the plans being offered are so unattractive. To entice people to join the exchanges, the administration forced insurers to offer low monthly premiums and cover people with preexisting conditions. Insurers have responded by increasing deductibles — the out-of-pocket costs people must pay before insurance benefits kick in — to stratospheric levels.

There must be a special place where people who predict hyperinflation in response to quantitative/credit easing, gold prices always going upward, and the collapse of ACA go to commiserate and convince themselves they are actually right.

Running List of Predictions and Predictors of ACA Collapse (send me your suggestions, so we can be as comprehensive as possible, and induce a little humility into these folks!)

Charles Krauthammer, October 28, 2013

Author: "Menzie Chinn" Tags: "health care"
Send by mail Print  Save  Delicious 
Date: Tuesday, 16 Sep 2014 03:53

Bruce Bartlett brought my attention to this article, which Mark Thoma mused was “The Stupidest Article Ever Published”. From The Inflation Debt Scam, by Paul Craig Roberts, Dave Kranzler and John Williams:

To understand how risky the rise of debt is, nominal debt must be compared to real GDP. Spin masters might dismiss this computation as comparing apples to oranges, but such a charge constitutes denial that the ratio of nominal debt to nominal GDP understates the wealth dilution caused by the government’s ability to issue and repay debt in nominal dollars. …

I’m not a spin master, and yet I cannot help but dismiss this calculation as exactly comparing apples to oranges.

Nominal debt divided by nominal GDP is expressed in years — essentially years worth of GDP necessary to pay off the debt. I can understand what this calculation yields. In contrast, nominal debt is in $, real GDP is in Ch.2009$/year, so nominal debt divided by real GDP is a number expressed in years times dollars per Ch.2009$.

The authors present this figure to buttress their case:

roberts_kranzler_williams

Source: Roberts, Kranzler and Williams, “The Inflation Debt Scam,” The International Economy (Summer 2014).

As far as I can tell, the article is merely an excuse for Williams to haul out the fully discredited “Shadowstats” one more time.

By the way, according to Shadowstats, the US economy has been shrinking nonstop since 2004-05, on a year-on-year basis…

So, if this is not the stupidest article ever, it is in the running (along with Don Luskin’s 2008 gem).

Author: "Menzie Chinn" Tags: "economic indicators"
Send by mail Print  Save  Delicious 
Date: Sunday, 14 Sep 2014 22:14

A growing number of observers are starting to conclude that we’re never going to see the rebound in growth rates that many people had anticipated as the U.S. recovers from the Great Recession. Here I comment on a new paper in which Northwestern Professor Robert Gordon explains the basis for his pessimism.

A streamlined version of Gordon’s analysis would decompose the growth rate of output as the sum of growth in productivity (output per hour worked), growth in average hours worked per employee, and growth of number of people working. We received a temporary boost to growth from the second factor as a result of a cyclical rebound in the average workweek since 2010. There might be a further modest contribution from this source if hours per worker get all the way back to their average value over 1988-2008. But more than half of any boost we might hope to see has already been realized at this point.

12-month average values for hours per worker, 1949:M1-2014:M8.  Data source: BLS.

12-month average values for hours per worker, 1949:M1-2014:M8. Data source: BLS.

We also have enjoyed a temporary boost in growth from the third term, an increase in the number of people working, as a result of the declining unemployment rate. But with the unemployment rate already back down to 6.1%, it again seems unlikely that further drops in the unemployment rate can make a major contribution to future U.S. growth rates.

Source: FRED.

Source: FRED.

An alternative source of rising total employment would be a recovery in the labor force participation rate. But the decline in this began well before the Great Recession, and seems largely due to demographic factors that are unlikely to be reversed.

Labor force participation rate (solid) and estimated trend component (dashed).  Source: Aaronson, et. al. (2014).

Labor force participation rate (solid) and estimated trend component (dashed). Source: Aaronson, et. al. (2014).

Thus a key factor in any prediction of faster U.S. GDP growth would seem to be based on an improvement in productivity.
Gordon writes:

Forecasters universally predict that actual real GDP growth will increase from the 2.1 percent average of the past five years to between 3.0 and 3.5 percent per year over the next two to three years. For output to grow at that rate without a decline in the unemployment rate to four or even three percent would require some combination of a much slower rate of decline in the labor-force participation rate or even a reversal toward an increasing LFPR, as well as a revival of labor productivity growth well above the 1.2 percent average growth rate of the past decade, not to mention the 0.6 percent average over the past four years.

Economists usually think of long-run productivity growth as primarily determined by technological innovation and capital improvements. But cyclical factors also have an effect on productivity that extends much longer than some might suppose. The graph below plots the average growth in U.S. productivity over 5-year intervals ending at each indicated date. It’s pretty clear that these intermediate-run averages are very strongly affected by the decline in productivity associated with economic recessions and the rebound in productivity usually associated with recoveries.

Average annual growth rate of nonfarm business sector real output per worked over 5-year periods ending at indicated dates, 1952:Q1-2014:Q2.  Data source: FRED.

Average annual growth rate of nonfarm business sector real output per worked over 5-year periods ending at indicated dates, 1952:Q1-2014:Q2. Data source: FRED.

A boost in aggregate demand can lead to higher productivity for reasons of simple accounting. Suppose you are operating a retail store and have the same number of employees today as yesterday. If more customers show up to buy today, your measured productivity (how much you sell divided by the number of hours your employees put in) is necessarily going to increase. If the demand stays high, you might find it necessary to hire more workers to meet the higher volume. But since many tasks (such as security, cleaning, and maintenance) won’t double when sales double, there will still be a gain in measured productivity when demand goes up. Something similar is true for many manufacturing facilities that have significant underutilized capacity.

The strong relation in the data between measured productivity and the strength of demand can be seen by plotting productivity together with other strongly cyclical series such as purchases of consumer durables. When consumers are trying to reduce spending, big-ticket items take the brunt of the cutbacks, and when confidence finally returns, purchases of consumer durables make up an unusually high fraction of total spending. Periods when productivity is growing slower than average correlate pretty strongly with episodes when consumer durables make up a smaller fraction of total spending than average. Durable purchases remain depressed today, and that may be a factor in continuing weak productivity numbers.

Top panel: productivity growth.  Bottom panel: spending on consumer durables as a percentage of GDP.  Horizontal lines represent historical averages.

Top panel: productivity growth. Bottom panel: spending on consumer durables as a percentage of GDP. Horizontal lines represent historical averages.

To be sure, many economists interpret the causation as running in the opposite direction. They argue that if technological improvements have been particularly favorable (so that productivity growth has been strong), consumers may become more optimistic about the future, and therefore devote a higher fraction of their budgets to big-ticket items. But to me, the natural interpretation of the correlation is the traditional Keynesian one that I described above as resulting from the direct effect of higher spending on measured productivity.

For this reason, I am a bit more optimistic than Gordon. At least as far as the next several years are concerned, I side with those who expect above-average growth as consumers’ and firms’ finances continue to recover. But Gordon makes the interesting and I think correct observation that some of the U.S. economic growth since 2010 has already come as a result of factors such as a lengthening workweek and falling unemployment rate, developments that by their nature cannot be projected to continue for much longer. I agree with Gordon that the U.S. should expect a slower growth rate of potential output over the next decade than we saw in most of the twentieth century. But that doesn’t mean that we still couldn’t see better growth numbers for 2014:Q2-2015:Q4 than we’ve experienced on average over the last five years.

Author: "James_Hamilton" Tags: "economic growth, economic indicators, em..."
Send by mail Print  Save  Delicious 
Date: Friday, 12 Sep 2014 22:32

How much can an independent Scotland rely upon? updated 9/16

The U.S. Energy Information Administration has just released its 2014 International Energy Outlook.

northseaoilprodn1

Figure 1: Norway and offshore UK Oil production in millions of barrels (black), North Sea oil production, actual through 2013 and forecast (blue), and high and low price forecasts (gray lines). Source: EIA, IEO 2014, Sept. 9, 2014, and EIA.

EIA does not report forecasts of the UK share of North Sea production. UK production accounted for about 33% of total Norway and UK production in 2013, down from 80% in 1982 (the EIA statistics only go back to 1980). See this Economist article for a graph of UK oil-equivalent production extending back to 1970.

Obviously, the revenue implications depend on the price of oil as much as the production levels. However, declining quantity of production seems the be in the cards, and in any case, production levels will be substantially below where they have been.

Update, 9/16, 11:35AM Pacific: Frequent commentator Steven Kopits presents his views on the outlook for UK North Sea oil production, and the implications for Scottish independence.

Author: "Menzie Chinn" Tags: "energy"
Send by mail Print  Save  Delicious 
Date: Friday, 12 Sep 2014 01:20

As Ronald Reagan once said (although he did mean to say “stubborn”)

Regarding the implications of optimal currency area theory and Scottish independence, Reader Patrick Sullivan continues his reign of error, trying to argue that Canada did just fine, just like a bank crisis-free Scotland in a currency union would:

Canada didn’t have a central bank until sometime in the 1930s, and had a less severe depression than the USA.

Well, with a little help from Louis Johnston, (who knows economic history much better than I), I generate the following plot:

can_us_percapita

Figure 1: Log per capita income in 1990 International dollars for Canada (blue) and United States (red), normalized to 1929=0. Source: Maddison and author’s calculations.

I dunno, but these seem to be comparable declines in output. So, yes, no central bank operating until 1935 [1] (p.22), but no, Canada suffers a pretty big shock (Canada on a de facto gold standard until 1931 (p.21)).

It constantly amazes me how people make easily falsifiable statements with such astounding confidence…

So, I think Scotland should consider very carefully independence conjoined with monetary union with England.

Author: "Menzie Chinn" Tags: "economic indicators"
Send by mail Print  Save  Delicious 
Date: Wednesday, 10 Sep 2014 01:51

A rapid collapse in the Wisconsin fiscal prospects (but pretty predictable, as long as one doesn’t believe in supply side miracles).

Three weeks ago, I documented the deterioration in Wisconsin budget prospects. Since then, the fiscal hemorrhaging has continued.

wibalance3

Figure 1: (Negative of) General Fund Amounts Necessary to Balance Budget, by Fiscal Year, in millions of dollars (blue bars); and estimate taking into account shortfall of $281 million for FY2013-14, and total $1.8 billion for biennium (red squares). “structural” denotes ongoing budget balance, assuming no revenue/outlay change associated with economic growth. Source: Legislative Fiscal Bureau (September 8, 2014), Wisconsin Budget Project, and author’s calculations.

The FY2013-14 estimate (red square) incorporates the estimated $281 tax revenue shortfall. It does not incorporate the likely higher than budgeted Medicaid expenditures.[Peacock/WBP] I estimate the FY2014-15 shortfall using the $1.8 billion estimated shortfall for the biennium noted by WisPolitics.

The downward revision in the structural budget balance since May is also remarkable.

wirevision

Figure 2: Change in the (Negative of) General Fund Amounts Necessary to Balance Budget, by Fiscal Year, in millions of dollars, going from May 22 to Sept 3 (red bars). Source: Legislative Fiscal Bureau (May 22, 2014), Table 6, and Legislative Fiscal Bureau (September 8, 2014) and author’s calculations.

But at least we gave tax cuts to upper income households!

Author: "Menzie Chinn" Tags: "budget, Wisconsin"
Send by mail Print  Save  Delicious 
Date: Tuesday, 09 Sep 2014 14:50

Ronald MacDonald (Glasgow) concludes that a currency union will not work for Scotland. From The Guardian:

One of the world’s top economists has warned that an independent Scotland’s economy would crash within seven years if it tried to use sterling.

Professor Ronald MacDonald, a currency expert who advises the International Monetary Fund and the European Central Bank, said the Scottish government’s plans to use sterling after a yes vote were fundamentally flawed, even if Alex Salmond’s proposals for a currency union were accepted by the UK. The Scottish economy would shrink by up to £100bn by 2023, MacDonald said.

For a more technical exposition, see this paper. Discussion of optimal currency area theory applied to the euro zone is here

See also the points laid out by Paul Krugman, focusing on the importance of fiscal federalism in currency zones.

Author: "Menzie Chinn" Tags: "international"
Send by mail Print  Save  Delicious 
Date: Tuesday, 09 Sep 2014 03:01

Newcomers to macro often encounter problems in interpreting and using data. The first is how to report growth rates, particularly when trying to assess the current state of the economy. The second is how to read data reported at annual vs. quarterly vs. monthly rates. The third is accounting for the presence of breaks in data collection. (This post primarily for students in Econ 435 and Public Affairs 854.)

Growth Rates

Suppose one wants to report annual growth rates, but one has a choice between using annual data or quarterly data. One convention applied to quarterly data is to use 4q/4q, that is the growth rate from the previous year’s 4th quarter to the current year’s. This calculation sometimes leads to similar growth rates, but sometimes does not. Consider a comparison of the years around the last recession.

yy_vs_qq

Figure 1: Annual GDP (blue bar), and quarterly GDP SAAR (red line), in bn. Ch.2009$, 2006-2009. Source: BEA.

The growth rate in 2008 using annual data is -0.29%, which seems like a modest decline. The 4q/4q change in 2008 was -2.77% (and the 2009Q4 2008Q4 q/q annualized growth rate was -8.35 -8.19%). Which calculation is correct? Neither — it depends on what the question is. However, if one were looking at the data in early 2009, one would probably want the most recent reading on GDP growth; hence a q4/q4 reading is probably more useful in this case.

Another way to look at this is that the year-on-year calculation is approximating a growth rate calculated from the middle of the previous year to the middle of the current year. Figure 2 shows the growth rates calculated using year-on-year annual data (blue), 4q/4q using quarterly data (red), and the 4 quarter change in the average of Q2 and Q3 GDP (green).

comparegrowth

Figure 2: Real GDP growth rates calculated using year-on-year annual data (blue line), 4q/4q using quarterly data (red line), and the 4 quarter change in the average of Q2 and Q3 GDP (green line), in %. Source: BEA and author’s calculations.

Bottom line: Year-on-year annual growth rates are essentially measuring the growth rates assessed from the middle of the year.

Annualized and Non-annualized Rates

Oftentimes, one sees the term SAAR ascribed to a data series; this abbreviation stands for Seasonally Adjusted at Annual Rates. Typical US government macro aggregates are typically reported at annual rates — for instance the GDP levels (red line) reported in Figure 1 are at an annual rate — that is the level of GDP that would obtain if that rate of spending in that quarter continued for an entire year.

This is something that even University of Chicago macroeconomists can make mistakes with. As I pointed out at the time, if one is calculating a multiplier as the change in GDP for a given dollar amount of stimulus (ΔY/ΔG), for instance, one has to remember the GDP figure is reported at annual rates, while the stimulus figure was (in this case) reported at a quarterly rate. Paul Krugman also noted that if one forgets that growth rates (in the US) are typically reported at annualized rates, one can make a compoundingly embarrassing error.

Just to confuse matters, not all US government series are reported at annual rates; some are not. For instance quarterly Balance of Payments data in the international transactions release are reported at quarterly rates; monthly data in the trade release are reported at monthly rates.

And once one starts looking at foreign data, one has to be particularly attentive; quarterly GDP level and growth rates are typically reported at quarterly rates.

Data Breaks

As government and international statistical agencies collect data, the mode of the data collection or the means of calculation sometimes change. Those changes are usually noted, but if one does not read the documentation, one can make serious mistakes. For instance, examining civilian employment (FRED series CE16OV), one would think there was a tremendous boom in employment in January 2000.

civempl

Figure 3: Civilian employment (blue, left scale) and change in civilian employment (red, right scale), both in 000’s, seasonally adjusted. Source: FRED series CE16OV.

If one inspects other series, say nonfarm payroll employment, one sees no corresponding jump. This suggests the boom interpretation is wrong. Further evidence of a “break” is found by inspecting the first difference of the series (in red). The spike in January 2000 is a 1.5% change (m/m), while the the standard deviation of changes is 0.3% (calculations in log terms).

In fact, the jump is due to the introduction of new population controls associated with the Census. New controls are applied every decade, so this is a recurring (and known — to those who are careful) problem. Nonetheless, here’s an example of the mistake regarding the participation figure. Other breaks are less obvious. This is a cautionary note to all who download data without consulting the documentation.

Update, 9/10:

Rounding Errors

In principle, real quantity = total value/price deflator. For instance GDP09 = GDP/PGDP09, where GDP is measured in $, PGDP09 is the GDP deflator taking on a value of 1 in 2009, and GDP09 is GDP measured on 2009$. In practice, there is a slight rounding error, which typically does not make a difference, but can if (1) growth rates are very high (or very negative), and (2) one is annualizing quarterly growth rates.

I used the manually deflated series for the 2008Q4 q/q calculation, when in this case it would have been better to use the real series reported by BEA to do the calculation.

Figures 4 and 5 show the rounding errors.

comparecalcreported

Figure 4: GDP in bn. Ch.2009$ SAAR, calculated by deflating nominal GDP with the GDP deflator (blue) and as reported by BEA (red), FRED series GDPC1. Source: BEA 2014Q2 2nd release and author’s calculations.

comparecalcreported1

Figure 5: Quarter-on-quarter annualized growth rate in GDP in bn. Ch.2009$ SAAR, calculated by deflating nominal GDP with the GDP deflator (blue) and as reported by BEA (red), FRED series GDPC1. Source: BEA 2014Q2 2nd release and author’s calculations.

Author: "Menzie Chinn" Tags: "economic indicators"
Send by mail Print  Save  Delicious 
Date: Sunday, 07 Sep 2014 18:10

Last week I commented on a puzzling phenomenon in bond markets this year– long-term rates have been falling at the same time that nearer-term rates have been rising. Bruegel has a review of some of the discussion of this around the web.

Nominal yields on different U.S. Treasury securities as a function of years to maturity as of January (green) and September (blue) of 2014.  Data source: Adrian, Crump, and Moench.

Nominal yields on different U.S. Treasury securities as a function of years to maturity as of January (green) and September (blue) of 2014. Data source: Adrian, Crump, and Moench.

The same shifts are seen in both the nominal yield curves (above) and the real yield curves (below). The interpretation I offered is that the higher near-term rates reflect the perception that the time when the Fed begins raising short-term rates (likely some time next year) is now less far away that it was in January, while growing pessimism about longer term growth prospects may account for falling longer term yields. However, this interpretation is difficult to reconcile with this year’s strong stock market.

Real yields on different U.S. Treasury Inflation Protected Securities as a function of years to maturity as of January (green) and September (blue) of 2014.  Data source: Gürkaynak, Sack, and Wright.

Real yields on different U.S. Treasury Inflation Protected Securities as a function of years to maturity as of January (green) and September (blue) of 2014. Data source: Gürkaynak, Sack, and Wright.

My interpretation assumed that the changing yield curve largely reflects changing expectations about future rates. But we know that another important component of the yield curve is the term premium, an added compensation you can expect to receive if you are willing to tie up your money long term. I provided a simple correction for term premia in my discussion last week. David Beckworth looked at some more sophisticated estimates of term premia developed by Adrian, Crump, and Moench. These suggest that term premia on all bonds have fallen significantly this year, with the changing 5-year 10-year differential largely explained by the fact that term premium on the 10-year bond has fallen more than that on the 5. However, this basically just passes the puzzle over to the unanswered question of why the term premium on longer-term bonds should have fallen so much more than that on short-term securities at the same time that the Fed is winding down its purchases of longer-term securities. Below I plot the contribution of term premia to the yield on each security as inferred by the Adrian, Crump, and Moench methodology.

Term premia on different U.S. Treasury securities as a function of yield to maturity as of January (green) and September (blue) of 2014.  Data source: Adrian, Crump, and Moench.

Term premia on different U.S. Treasury securities as a function of yield to maturity as of January (green) and September (blue) of 2014. Data source: Adrian, Crump, and Moench.

Unfortunately, separating the contribution of term premia is not that easy to do. The basic idea is to form a model to form a good forecast of future short-term rates, and interpret the difference between the observed bond yield and the discounted value of the model’s expected future short rates as the term premium. But different reasonable approaches to forming those forecasts can result in very different assessments of movements in term premia. The figure below shows that various approaches can differ wildly in terms of the answer they give.

Term premium component of the yield on 10-year Treasury security as inferred from 4 different methods.  Source: Swanson (2007).

Term premium component of the yield on 10-year Treasury security as inferred from 4 different methods. Source: Swanson (2007).

Econbrowser reader JBH also called attention to the fact that the yield on long-term German bonds has fallen even more dramatically this year than U.S. Treasuries, suggesting that some of the source for the flattening U.S. yield curve could be global.

Yields on 10-year U.S. Treasuries (black) and long-term German bonds (blue), monthly, Jan 2001 to July 2014.  Data source: ECB.

Yields on 10-year U.S. Treasuries (black) and long-term German bonds (blue), monthly, Jan 2001 to July 2014. Data source: ECB.

Robin Harding and Michael McKenzie note in the Financial Times how unusual this year’s divergence between the U.S. 5-year and 10-year rates is. I did a calculation related to theirs, looking at the correlation between monthly changes in the 5- and 10-year rates over 12-month periods. This correlation today is the lowest it’s been in the last 30 years.

Zero-mean correlation over 12-month period ending at each date between monthly change in 5-year and 10-year rates.

Zero-mean correlation over 12-month period ending at each date between monthly change in 5-year and 10-year rates.

On the other hand, the co-movement between U.S. and German yields this year has not been that unusual. For example, in 2011, the two moved together much more closely than they have this year.

Zero-mean correlation over 12-month period ending at each date between monthly change in yield on 10-year U.S. Treasury bond and that on long-term German bond.

Zero-mean correlation over 12-month period ending at each date between monthly change in yield on 10-year U.S. Treasury bond and that on long-term German bond.

To the extent that Europe’s long-term prospects have dimmed, perhaps investors are willing to accept a lower real return on long-term investments in the United States, and this may account for part of the puzzle. And if the market expects the Fed to start raising rates sooner than the weak world economy warrants, that may account for some of the unusual behavior of the 5-year rate relative to the 10-year rate.

Author: "James_Hamilton" Tags: "Federal Reserve, financial markets, inte..."
Send by mail Print  Save  Delicious 
Date: Friday, 05 Sep 2014 03:33

From Simon Kennedy in “Draghi Sees Almost $1 Trillion Stimulus With No QE Fight” (Bloomberg):

Mario Draghi signaled at least 700 billion euros ($906 billion) of fresh aid for his moribund economy and left a fight with Germany over sovereign-bond purchases for another day.

Draghi gave few details on the size or nature of the ABS plan, saying the “modalities” of the program will be announced in October. The ECB will initially target the less-risky segment of the market, which collapsed in the wake of the financial crisis after being criticized by politicians and regulators for being opaque. The ECB may consider buying more-risky versions if governments provide a guarantee.

“We want to make sure that these ABS are being used to extend credit to the real economy,” Draghi said. The measures “are predominantly oriented to credit easing.”

Weidmann, who has clashed with Draghi before over the need for stimulus, opposed the policy measures, according to two euro-region central bank officials. A Bundesbank spokesman declined to comment. In July, the German central-bank head called ABS purchases “problematic” and warned against supporting bank profits while socializing the losses.

Commerzbank AG economist Joerg Kraemer said there is still a 60 percent probability the ECB will eventually buy government bonds should its forecasts deteriorate.

One measure of the degree to which this announcement signals a looser monetary policy is indicated by the intra-day movement in exchange rates.

dollar_euro_5day

Figure 1: USD/EUR exchange rate, last five days. Down is a euro depreciation. Source: Reuters.

Context on the magnitude of the move is provided by reference to a longer time span — in this case, year to date.

dollar_euro_exrate

Figure 2: USD/EUR exchange rate, year to date. Down is a euro depreciation. Source: FRED and Pacific Exchange Services.

The drop against the dollar day-on-day was 1.6% (log terms), where the standard deviation of changes during the year-to-date was 0.3%. Update 9/7: this depreciation shows up in the trade weighted exchange rate as well.

ea_eer


Figure 2a: Trade weighted value of the euro, over the last year. Source: ECB.

All I can say is that this measure is long overdue, given the sad state of Euro area economic growth, and the troubling deceleration in inflation, now at 0.3%.

ecoin_24Jan13_word

Figure 3: Quarter on quarter nonannualized Euro area GDP growth (red dots) and Euro coin indicator (blue line). Source: Eurocoin/CEPR.

hicpinflation1

Figure 4: Year on year HICP inflation in the euro area (blue), and 2% ECB inflation target (red). CEPR defined recession dates shaded gray. Dashed line at mid-quarter of last announced peak (2011Q3). Source: FRED, CEPR, author’s calculations..

Update, 9/6 12:10PM: From Torsten Slok (9/5, not online):

The ECB wants to expand its balance sheet by EUR1trn, but the question is if they can find enough ABS and covered bonds to buy, see also the chart below. The direct impact on economic activity of balance sheet expansion is very limited. Instead, the main transmission channel of monetary policy here is via a depreciation of the euro. Expect EUR/USD to continue to grind lower, this is good for European exports and for European inflation and hence also for the ECB.

ecb_abs

Author: "Menzie Chinn" Tags: "exchange rates, financial markets, inter..."
Send by mail Print  Save  Delicious 
Date: Wednesday, 03 Sep 2014 03:44

As I begin teaching finance in the new semester, I am highlighting the key puzzle of our times, discussed by Jim in his last post, with this graph:

debt_intrates

Figure 1: Ratio of Federal debt held by public to GDP (bold black, left scale), ten year constant maturity Treasury yield (blue, right scale), ten year Treasury yields minus ten year median expected inflation (green, right scale), and TIPS ten year constant maturity yields (red, right scale). NBER defined recession dates shaded gray. Source: Treasury, Federal Reserve Board, BEA (2014Q2 2nd release), Survey of Professional Forecasters, NBER, and author’s calculations.

The key point of this graph is that the US debt-to-GDP ratio has risen, starting upward in 2008Q3; and yet nominal ten year yields have continued to decline, despite the warnings of Paul Ryan, and others, contra the crowding out hypothesis. Some of this decline can attributed to a decline in expected inflation, but not by any means most. Real interest rates calculated using yields minus expected inflation have also declined. Aside from the financial turmoil surround Lehman’s collapse, TIPS yields confirm the decline.

Why have government yields declined so much, despite the reversal in the debt-GDP ratio engineered by the Bush Administration (remember EGTRRA and JGTRRA?). About ten years ago, Jeff Frankel and I started investigating this puzzle [1]. We regressed US yields on actual and expected US debt-GDP ratios (annual data), and found that up to 2002, the canonical regression worked well. But incorporating data up to 2006, the correlation broke down — and yields were over-predicted. This finding was one manifestation of the “conundrum”. Only by incorporating foreign purchases of US Treasurys could we restore the correlation (and fit the data).

So, looking forward, what can we expect to happen to yields, given foreign and Fed holdings of US Treasurys? First a regression over the 1991Q4 to 2002Q4 period.

rt = 0.026 + 0.071×dt -0.348×fedt – 0.067×fort + 0.253×yt-1

Adj.-R2 0.34, SER = 0.005, DW=0.91, n=45. bold face denotes significance at the 10% MSL, using HAC robust standard errors. Where r is the ten year real interest rate, d is debt-to-GDP, for is foreign and international holdings of Treasurys and Fed is holdings by the Federal Reserve Banks (both normalized by GDP), and y is the output gap, measured using the log of the ratio of GDP to the CBO estimate of potential GDP (February 2014 release).

For each 1 percentage point increase in the debt-to-GDP ratio, the Treasury yield rises by 0.07 percentage points. This estimate is similar to the figure obtained in Chinn and Frankel for the period 1988-2006. Interestingly, the coefficients on foreign and Fed holdings are not statistically significant.

Expanding the sample to 2014Q1, one obtains the following results:

rt = 0.023 + 0.073×dt -0.159×fedt0.130×fort + 0.223×yt-1

Adj.-R2 0.83, SER = 0.005, DW=0.76, n=90. bold face denotes significance at the 10% MSL, using HAC robust standard errors.

Including a time trend does not change the results (and the coefficient on the time trend is not significant in itself). Interestingly, the hypothesis that for and Fed are equal and opposite sign from that on d cannot be rejected at conventional levels (similar to Kitchen and Chinn (2012)).

The fit is much better using the full specification, as opposed to merely debt and output gap. This is shown in Figure 2.

yield_fit

Figure 2: Ten year Treasury yields minus ten year median expected inflation (bold green), fitted values using full specification (dark blue), and using partial specification (purple). NBER defined recession dates shaded gray. Source: Federal Reserve, Survey of Professional Forecasters, NBER, and author’s calculations.

(Note: it would be preferable to use foreign official sector holdings, as opposed to foreign (private and public) holdings, but this is the series I could get on quick notice.)

While Fed holdings will stabilize and shrink over time, it’s not clear to me that foreign holdings expressed as a share of US GDP are going to shrink.

debt_held_intl

Figure 3: US Treasury debt held by foreign and international sector, in billions $ (blue, left scale), and as share of GDP (red, right scale). Source: Flow of funds, BEA, and author’s calculations.

In fact, in the short term, one can imagine elevated holdings of US Treasurys, given geopolitical concerns. Over the longer term, the demand for “safe assets” (see definition here and here) will likely continue, as financial development is a very slow process (and regressing in places like Russia).

In sum, I am not so sure the case for a rebound of long term rates to the norms of the post War era is so strong.

Update, 9/3 9:15AM Pacific: See also DeLong/WCEG, who brings my attention to the blog review by Cohen-Setton/Bruegel.

Author: "Menzie Chinn" Tags: "deficits, Federal Reserve, financial mar..."
Send by mail Print  Save  Delicious 
Date: Sunday, 31 Aug 2014 19:12

As the U.S. economy returns to healthier growth, many of us expected long-term interest rates to return to more normal historical levels. But the general trend has been down since the end of the Great Recession. The 10-year rate did jump back up in the spring of 2013. But during most of this year it has been falling again.


Nominal yield on 10-year U.S. Treasury bond, from FRED.

Nominal yield on 10-year U.S. Treasury bond, from FRED.

And it’s not just that people are getting used to the idea that inflation is always going to be low. The drop in the yield on a 10-year Treasury inflation-protected security (TIPS), whose coupon and par value go up with the headline CPI, has also been impressive.

Yield on 10-year TIPS.  Data source: Gürkaynak, Sack, and Wright.

Yield on 10-year TIPS. Data source: Gürkaynak, Sack, and Wright.

But here’s an interesting detail. While the return on a 10-year Treasury has been falling for most of this year, the 5-year yield has held fairly steady.

Nominal yield on 5-year (blue) and 10-year (black) U.S. Treasury bonds.

Nominal yield on 5-year (blue) and 10-year (black) U.S. Treasury bonds.

Sometimes we summarize the relation between those two yields using a forward rate. If today you simultaneously bought $1000 worth of a pure-discount 10-year bond and sold $1000 worth of the 5-year discount bond, it would be a complete wash in terms of the cash flow for the next 5 years. Five years from now you’d have to pay back the 5-year bond, and 10 years from now you’d get the proceeds from the 10 year bond. In effect you’ve locked in the terms today on a 5-year bond that you’re not going to buy until 5 years from now. The yield on that bond is known as the 5-year-5-year forward rate, and is plotted below.

Five-year-five-year forward rate.  Calculated as 2 times the 10-year rate minus the 5-year rate.

Five-year-five-year forward rate. Calculated as 2 times the 10-year rate minus the 5-year rate.

If investors were risk neutral, you could read that forward rate as the market’s expectation of where the 5-year rate is going to be 5 years from now. The drop in the forward rate during 2014 would indicate that something happened this year to persuade people that rates in the future were going to be lower than they had been expecting.

But there’s lots of evidence that markets are not risk neutral. The 10-year rate is almost always higher than the 5-year rate, not because people think that the 5-year rate is going to increase, but because they want extra compensation for locking up their money longer term. I made a very quick adjustment for risk aversion by using data since 1990 to regress the actual 5-year rate on a constant and what the 5-year-5-year forward rate had been 5 years previously, and used the fitted values of that regression as a “risk-adjusted” forward rate. This adjusted series is shown in blue along with the original in black in the graph below. The message is unchanged.

Five-year-five-year nominal forward rate (black) and forward rate adjusted for risk premium (blue).

Five-year-five-year nominal forward rate (black) and forward rate adjusted for risk premium (blue).

You can also calculate a real 5-year-5-year forward rate using the TIPS. Once you correct for pricing of risk, it looks like investors have been anticipating a negative real rate well into the future ever since the end of the recession. Expectations lifted in the first half of 2013, but have been falling sharply this year.

Five-year-five-year real forward rate (black) and forward rate adjusted for risk premium (blue).

Five-year-five-year real forward rate (black) and forward rate adjusted for risk premium (blue).

What could produce such a pattern? It’s hard to attribute it to changing perceptions about the Fed, which should surely matter more for the next 5 years than they would for 5 to 10 years from now. More confidence that the U.S. government will be able to keep debt from growing relative to GDP over the next decade may have played a role.

Another possibility is that more people are starting to take seriously the suggestion that we’re on a path now of secular stagnation with weak economic growth and poor investment opportunities over the next decade. But that’s hard to reconcile with the stock market, which climbed impressively this year.

S&P 500.  Source: Google Finance.

S&P 500. Source: Google Finance.

Or then again, maybe the market has simply overpriced both stocks and long-term bonds.

Author: "James_Hamilton" Tags: "Federal Reserve, financial markets"
Send by mail Print  Save  Delicious 
Date: Saturday, 30 Aug 2014 21:07

No succor from the QCEW series that the Walker Administration previously touted [1]

DWD released QCEW figures for first quarter 2014. Since the QCEW figures are not seasonally adjusted, I have estimated a series consistent with the BLS private nonfarm payroll employment series based on the QCEW data.

wijulpix

Figure 1: Wisconsin private nonfarm payroll employment (blue), path implied by Governor Walker’s August 2013 pledge (red), and employment implied by QCEW series (orange), all in thousands, seasonally adjusted. Source: BLS, DWD, and author’s calculations.

Figure 1 includes an estimate of the private nonfarm payroll series using the QCEW figures as input. By March 2014, the BLS series and the estimated are close, although the BLS establishment survey was higher for much of the preceding months.

This result confirms Wisconsin’s lackluster employment growth. As noted in this post, in order to hit the January 2015 target laid out by Governor Walker, the Wisconsin economy will need to generate 22.6 thousand net new jobs in each of every month until January. Mean job creation over Governor Walker’s term thus far has been 2.7 thousand per month. This suggests that it is unlikely that the goal will be achieved.

The implied employment level is obtained by regressing the log BLS series on the log QCEW series, along with a constant and monthly seasonal dummies, over the 2001M01-2014M03 sample. The elasticity is 0.96, the adjusted-R2 = 0.99, and SER = 0.0019 (much as in the earlier case, suggesting stability in the relationship).

In other news, “… tax collections were $281 million less than anticipated for the fiscal year that ended in June. That puts the two-year budget on pace to be at a $115 million shortfall by June 30.” In other words, the hit to tax revenues was $81 million more than I assumed in this post. See this for additional information regarding the full enormity of the shortfall.

Author: "Menzie Chinn" Tags: "Wisconsin"
Send by mail Print  Save  Delicious 
Date: Friday, 29 Aug 2014 20:55

And Kansas travels its own path

Bruce Bartlett brings my attention to this article noting Minnesota’s economic performance. This reminded me to check on the Philadelphia Fed’s forecast for the next six months, released earlier today. What’s interesting to me is the fact the cumulative growth gap between Minnesota and Wisconsin (relative to 2011M01) is forecasted to grow — rather than shrink — over the next six months.

states_jul14

Figure 1: Log coincident indices for Minnesota (blue bold), Wisconsin (red bold), Kansas (green), California (teal), and United States (black), all normalized to 2011M01=0, seasonally adjusted. Observations for 2015M01 are forecasts implied by leading indices. Source: Philadelphia Fed and author’s calculations.

For those who argue that because Minnesota fell further during the Great Recession so it should be growing faster than Wisconsin — well that’s just plain wrong (not that I believed in that particular snapback argument). Wisconsin fell 8.3% from 2007M12 to trough, while Minnesota fell 5.0% (both in log terms).

The cumulative growth gap between Kansas and the Nation is also forecasted to rise, from the current gap of 2.7%, to 3.2%, in just the next six months. The forecast from a simple ARIMA(1,1,1) estimated over the 1986M01-2014M07 period yields the same conclusion: the gap will widen.

Author: "Menzie Chinn" Tags: "Wisconsin"
Send by mail Print  Save  Delicious 
Date: Friday, 29 Aug 2014 17:11

From Bloomberg::

The U.K. will press European Union leaders to consider blocking Russian access to the SWIFT banking transaction system under an expansion of sanctions over the conflict in Ukraine, a British government official said.


“There’s no doubt that in the short term restricting Russian usage of SWIFT would be extremely disruptive to Russian financial and commercial activities,” said Richard Reid, a research fellow for finance and regulation at the University of Dundee in Scotland.

Russian economic authorities themselves accede to the increasing likelihood of recession [0], just as non-Russian forecasters [1]. Economists’s beliefs that no further sanctions would be imposed, cited in the Bloomberg survey, are likely to be tested by new information regarding the incursion of Russian forces in Ukraine. (Note: the ruble has hit a new low, despite likely forex intervention; MICEX again down, 1.32%, RTS cash index, down 2.11%).

It seems no longer so implausible that the worst case scenario for the Russian economy, including a cumulative 5% decline in output (IIF, March 2014), becomes a reality.

russia-gdp-growth

Figure 1: Russian real GDP q/q growth rate (non-annualized). Source: TradingEconomics. Last observation is for 2014Q1.

Author: "Menzie Chinn" Tags: "international"
Send by mail Print  Save  Delicious 
Date: Thursday, 28 Aug 2014 20:16

Edit/update 8/30:

T72BM_in_Ukraine

T72BM (only known to be in service in the Russian army) in Ukraine, 8/26. Source: IISS


From NYT:

NATO released photographs on Thursday that it said shows Russian artillery units operating in Ukraine, and asserted that more than 1,000 Russian soldiers had now joined the separatists fighting there against the Ukrainian armed forces.

29UKRAINE-2-articleLarge

Source: NYT.

The MICEX is down 1.84% today; the Russian Trading System Cash Index is down 3.43%.

image001

Source: Bloomberg.

Thus far, risk indicators such as VIX are not evidencing upward spikes, but that could change very quickly.

Given the downturn in Western Europe — partly attributable in Germany to the slowdown in exports to Russia [1] — this is unwelcome news for those who are looking forward to recovery in Europe. This development certainly places greater emphasis on expansionary monetary policy, and relaxation of contractionary fiscal policies.

Furthermore, the fairly dire predictions for the Russian economy discussed in this post seem ever more likely to be realized. [2]

Update, 8/30 2:45PM Pacific: Additional information regarding Russian supply of main battle tanks to the separatists, from IISS

Author: "Menzie Chinn" Tags: "international"
Send by mail Print  Save  Delicious 
Next page
» You can also retrieve older items : Read
» © All content and copyrights belong to their respective authors.«
» © FeedShow - Online RSS Feeds Reader