The CERF Blog

Goodhart’s Law (named after economist Charles Goodhart) says that once an observed empirical relationship begins to be relied upon, it will no longer work.   This law, or a variant, comes up in a lot of fields, but especially finance and economics.  Consider the problem of forecasting the rate of inflation.    Price stability is a major objective of monetary policy.  Policy actions take effect over time (“long and variable lags”), therefore estimation of where inflation will be one to two years out is key to successful policy.  However, accurate forecasting has proven difficult in practice.  In particular, researchers James Stock and Mark Watson¹ have examined the effectiveness of inflation forecasting rules based on various indicators; indicators like commodity prices, wage growth, money supply, and have found that no one of them consistently does better than a simple naïve rule, such as assuming next period’s inflation will be the same as today’s. 

Does this mean that economic forecasters are dropping the ball?  No, it is simply an application of Goodhart’s Law.  To see this, suppose that reliable leading indicator for inflation was discovered based on, let’s say, the price of anchovies.  Once economists at the Federal Reserve became aware of this rule – that an increase in the price of anchovies today translated into a change in the overall inflation rate next quarter – then the Fed would tighten policy when the price of anchovies went up, preventing the overall inflation and killing the usefulness of the rule.

A real world example arose in the 1970s when the Federal Reserve, relying on the excellent historical correlation of money growth and inflation, decided to target the rate of money growth in order to control inflation.  But just about the time that the Fed adopted a money growth target, the stability of the historical relationship between inflation and money growth dissolved. 

One way to attempt to overcome Goodhart’s Law is by focusing directly on your objective instead of a proxy.  When you focus on a proxy, the assumed relationship between objective and proxy is subject to change, and the proxy is subject to being gamed.  A famous example of the latter problem was the directive issued by USSR Central Planning.  When nail producers were rewarded based on the weight of nails delivered, a small number of giant nails were forthcoming.  When the reward was changed to the number of nails, millions of very small nails were delivered. 

The mandate of the European Central Bank (ECB) is low and stable inflation.  Thus, for the ECB to explicitly target an inflation rate (1.5-2%) serves as an antidote to Goodhart’s Law.  The U.S. Federal Reserve has a dual mandate, to foster both price stability and full employment.  Some economists, including FOMC members such as Ben Bernanke, have argued that the Fed should adopt an explicit inflation target.  Others fear that setting an explicit target would create the impression that the Fed was not following its dual mandate.

A better approach may be to target nominal GDP (NGDP).  Targeting NGDP automatically addresses both halves of the dual mandate.  Also, proponents argue that Fed policy would be less unstable.  In particular, NGDP targeting would call for lesser monetary policy adjustment to supply shocks than does inflation targeting.  For example, consider a negative supply shock, like at oil price spike for example.  This would tend to drive output down and measured inflation up.  If the Fed were focused on inflation then policy tightening would be in order.  This would make the output decline more severe.  Conversely, given a positive supply shock, like an increase in productivity, output would be greater and the price level lower.  Price targeting would call for an ease in policy, which could easily support asset price bubbles.  In both cases, NGDP targeting would call for lesser monetary adjustment.  Thus, in the face of a positive supply shock, NGDP targeting would be less likely to create an asset bubble, and in the face of a negative supply shock, NGDP targeting would be less likely to exacerbate the downturn.

In their meeting November 1-2, 2011, FOMC committee members discussed the merits of stating explicit targets for the price level or for NGDP.  The minutes refer to staff studies that suggested NGDP targeting could in principle be useful in supporting a stronger recovery.  However, there were a number of negative arguments raised including possible loss of anti-inflation credibility, the difficulty of setting appropriate targets and uncertainty created by making a change in policy.

These considerations are valid but not overwhelming.  I suspect that the Federal Reserve staff and FOMC will revisit this idea in the future.  By explicitly targeting the objective, the negative effects of Goodhart’s Law could be mitigated.  Sure, the connection between Fed policy instruments (e.g., the monetary base) and NGDP is subject to substantial variation.  But the essence of NGDP targeting is to make adjustments for such variation, the target being NGDP not the monetary base.

¹Stock, James and Mark Watson, “Forecasting Inflation,” Journal of Monetary Economics, 1999.

I’m often asked about how regional forecasts are constructed.  The fact is that different outfits construct regional forecasts differently.

At the lowest level of rigor, a forecaster may just discuss broad trends, perhaps referring to historical data, but not provide a numerical forecast.  These forecasts are not without value.  Improved understanding the general sweep of things can help forecast users determine the major influences on their organization.  Still, the information in this type of forecast is limited.

A slightly more rigorous forecaster may write down forecasted numbers, but the model generating the numbers is not formal, existing only in the forecasters mind.  Again, there is information in this forecast. The forecaster could be astute and very well informed.  The limitation is the lack of a structural model may result in inconsistencies, and explaining the forecast can be challenging.

The next level of rigor is a formal statistical model.  This has the advantage of clearly identifying links and causality, which is valuable information.

There are a few outfits that will provide a statistical forecast for every county in the United States, but there are problems with these forecasts.  One problem is that these models assume that links and causal relationships are identical everywhere.  The better models will weight links differently, but they assume the same links everywhere.  Also, because different data is available different counties and states, these models are limited by the data available for the most data-poor county.

A custom statistical model for a geography provides an improvement in information and rigor.  However, even this type of model is not the best that can be done.  These models can be constructed by analyzing data without ever visiting the region.  However, visiting the region provides information that enables a forecaster to construct a better model.  Regular visits provide opportunity to regularly improve the model.

Another, very serious, problem with purely statistical models is that these models are backward looking.  I like to use Shasta County California to illustrate this problem.  Suppose you have a statistical model of Shasta County California, but you don’t know that they found an owl in the woods, and the county’s lumber business will soon grind to a halt.  You will have a bad forecast, because the purely statistical model can’t use this information.  It only knows what has happened in the past, and even that with quite a lag for some types of data.

The most valuable type of model is a statistical model that has the ability to use information not yet reflected in the data.  This is the type of modeling we do, through the use of what the industry calls add factors.

I’ll use Central Oregon as an example:

I visited Central Oregon for the first time about seven months before our first forecast.  That visit was followed by two more visits.  Another economist accompanied me on one of those follow-up visits.  The purpose of these visits was to really understand Central Oregon’s economy, and that helped us develop a model that has performed pretty well.

Of course, building a model is not a one-time deal.  We are constantly trying to improve our models.  We also have to keep tabs on the economies we forecast.  So, continuing the Oregon example, we make several trips a year to Central Oregon.

We do more than just visit a region a few times a year.  We collect a lot of data for every region we forecast.  Maintaining and cleaning these data is an ongoing process.  We also follow a region through news and other media.  For example, we subscribe to the Cascade Business News in Central Oregon.

Technology helps us too.  We use Google to serve as a mail service.  We receive e-mail links to every article or blog on the State, region, County, and cities on topics such as jobs and economy.  We also receive e-mail links to articles and blog on the legislature.  Going through those e-mails and links is part of our everyday scan for every region we forecast.

We also maintain relationships with citizens of the regions we forecast.  These people will send us an e-mail or give us a call to tell us about events that they think could impact our forecast.  Consequently, it is not unusual for us to know about an event before it is public.

We also have a State forecast for every region we forecast.  You just can’t do a regional forecast without a State forecast.  The State forecasts are performed quarterly, and like a regional forecast, these require constant monitoring.  Sometimes, we have to do forecasts of other regions that may impact the region of interest.  For example, a Ventura County forecast has to include a Los Angeles forecast, if it is to be accurate.  Los Angeles County’s economy has a huge influence on Ventura County’s forecast.  Similarly, a Central Oregon forecast requires a current California forecast.

There you have it, a glimpse into our sausage machine.

The No Child Left Behind Act became law in 2002. Among other things, it required standardized testing of students, beginning in 2003. The scores are used to evaluate the quality of the schools.

It sounds reasonable. Congress certainly thought so. It was co-authored in the Senate by Edward Kennedy (D-MA) and Judd Gregg (R-NH), while John Boehner (R-OH) and George Miller (D-CA) introduced it into the House. It passed both houses by huge bi-partisan majorities, 91-8 in the Senate and 384-45 in the House.

The Act’s passage also marked the low point in California’s High School dropout rate.

In 2002, California’s High School dropout rate had been declining for several years. After the act’s passage, the dropout rate trend experienced an unprecedented reversal. What had been a declining trend became an increasing trend, one that continues today. After bottoming out at less than 11 percent in 2002, California’s High School dropout rate is now approaching 22 percent.

The costs of dropouts are enormous, both for the students who leave school and for society. A person without a High School education is economically crippled. For all but the very exceptional few, dropping out of High School is a sentence to a lifetime of poverty and drudgery. For many dropouts, a lifetime of poverty and drudgery is the best possible outcome. Far too many will be involved in drug abuse, dysfunctional or violent relationships, teenage pregnancies, and crime.

The costs to society are large. They include losses to crime, and the direct costs of subsidies, social programs, healthcare, prisons, and law enforcement. Those costs may be exceeded by the dropout’s output deficiency, that is, the difference between what the dropout would have produced with a decent education and what he or she actually produces.

One way to improve standardized test scores is to increase the retention of tested topics by the students. An easier way is to prohibit students who would perform poorly from taking the test. Since all students have to take the test, this means converting poorly-performing students into non-students, letting them drop out.

It looks to me like California’s educational establishment has opted for the easy way.

On the chart below, the purple line shows California’s dropout rate from 1997 through 2009; you can see the percentages on the right-hand side of the chart. The other lines show the percentage — on the left side of the chart — of California’s students who passed the standardized tests for Math, Language, and Science. California’s passing percentage in each field has increased lockstep as dropouts increased.

It is worse than that, though. The percentage of students passing the standardized tests has increased by about 15 percent, on average, while the percentage of students dropping out has just about doubled. That’s an extraordinarily expensive improvement.

Did the schools follow this strategy deliberately? You can’t rule it out. People react to incentives, and the Act provides an incentive to abandon those who will likely perform poorly on the tests. Teachers will probably object to that, but we have no reason to believe that they should somehow be different that most people and ignore the incentives. Besides, we’ve already seen examples of teachers and administrators cheating on these tests.

Teachers assert that the solution to all of No Child Left Behind problems is to abandon it. The other solution, of course, is to fix the incentives. The way to do that would be to assign the schools a huge financial penalty for dropouts. Teachers and administrators would scream. They would tell us that dropouts result from problems at home and socioeconomic conditions.

No doubt, many students have terrible home conditions that put these children at a huge disadvantage, but those are exactly the children that we should be giving the most attention. A lousy home environment doesn’t explain the sudden increase in dropouts. These issues have been with us for a very long time. I took my first college economics class, The Economics of Poverty, in the 1969-1970 school year. There is nothing about poverty today that we didn’t discuss in that class, except that the returns to education have increased dramatically since then.

Failure to educate disadvantaged children guarantees that the perverse cycle of poverty and despair is perpetuated. Providing them with quality education, even with the active resistance of family, friends, culture, and the students themselves, is the only way to provide them with even the minimum hope for the upward mobility that government-provided education implicitly promises.

Abandoning our least advantaged children is unconscionable. If we are to have an egalitarian and merit-based society, we must reduce the dropout rate. The way to ensure that no one is abandoned is to penalize the school for dropouts. It sounds harsh, but we owe it to the students, and we owe it to ourselves.

Previously appeared at newgeography.com

The Taylor Rule

The Taylor Rule relates the target federal funds rate with the gap between actual and target inflation and between actual and capacity output.  The higher is inflation or the lower is the output gap, the higher is the target funds rate.  The rule is both descriptive and prescriptive.  It was originally proposed by John Taylor in 1993 to describe or explain observed Fed behavior over the period 1986-1993.  For the most part, this was a period in which the Fed was targeting the funds rates.  Subsequently, the rule has sometimes been used to suggest what the Fed should do, and deviations between the funds rate and the target rate implied by the rule have been taken to indicate excessive Fed accommodation or lack thereof.  In particular, John Taylor himself has used this approach to argue that Fed policy was too easy in the 2002-2005 timeframe thus leading to the housing bubble.  Similarly, according to the Taylor Rule the Fed was too stimulative during the 1970s, thus leading to the “Great Inflation” of that era.

If you apply the Taylor Rule today, you are likely (subject to some equivocation due to difficulties of precisely measuring the inputs to the rule, thus enabling varying estimates of the target funds rate) to find a target rate below zero.  The original Taylor Rule is

            FFR = RR + PT + 1.5*(P-PT) + 0.5*Output Gap

            FFR = Target federal funds rate

            RR = Real rate of interest

            PT = Target inflation

            P = Actual inflation

            Output Gap = Percent deviation of actual output from potential output

Assuming a real rate of 1%, target inflation of 2%, an inflation gap of zero and an output gap of negative 9%, we obtain a target funds rate of negative 1.5%.  It is obviously difficult to target a funds rate below zero so, to the extent Fed policy attempted to target the funds rate, there would be a problem (this problem is known as the Zero Lower Bound (“ZLB”)).  The Fed believes that monetary policy can still be effective at the ZLB, even though the funds rate cannot be pushed lower. 

The fed funds target rate today is the range 0-.25%.  Since the observed rate is higher than the Taylor Rule, Fed policy could be viewed as “tight” today.  At least, it could be if it weren’t for Quantitative Easing (QE).  The Fed has been aggressively conducting QE.  In the past three years the monetary base has just about tripled.  Has the Fed done enough or too much? 

The McCallum Rule

To answer this question, we need a rule which is not constrained by the ZLB.  A good choice for this is the McCallum Rule named after, appropriately enough, economist Ben McCallum.  The McCallum Rule describes a path for growth in the monetary base that is consistent with achieving a target path for nominal GDP.  Suppose we target 5% growth in nominal GDP and attempt to define the path of base growth that would achieve this rate of nominal growth.  The linkage between nominal GDP and the monetary base is the velocity of the monetary base, that is, the ratio of nominal GDP to the base.  While the Fed can control the monetary base fairly closely, in order to achieve a nominal GDP target the Fed would have to adjust for changes in the monetary base.

Since 2007, base velocity has declined almost two-thirds (from 16 to 6).  Base velocity began to fall sharply in late 2008, coincident with the collapse in real and nominal GDP growth.  The Fed did react in real time, with monetary base growing by more than $700 billion in Q4 2008.  Is this consistent with the McCallum Rule?

The McCallum Rule¹ can be expressed as follows:

Base growth = 5.0% – Estimated Velocity Growth – k*(GDP Gap)

            Base growth = Growth rate of monetary base

            GDP Gap = Percent difference of nominal GDP from target GDP

Here we are assuming that 5% is a reasonable target growth rate for nominal GDP based on target inflation of 2% and a long-term trend real growth rate of 3%.  The GAP Gap is currently approximately minus 9%, similar to the Output Gap mentioned above.   The parameter k determines the speed at which the GDP Gap is closed.

Implementation of the McCallum Rule depends on the choice of the parameter k and the method used to adjust for changes in base velocity.  In our simulations, we have used a three-month moving average (that is, assume base growth next period is equal to the average growth over the prior three months).  We find that a value of k=.25 (for a quarterly simulation) describes a path for the monetary base over the past four years that is consistent with observed base growth (that is, the Bernanke Fed has effectively been operating according to a McCallum Rule with k=.25).  Values of k greater than .25 would have been more expansive that has occurred and values less than .25 would have been more restrictive than has been observed.

Nominal GDP Targeting

Even if base velocity stabilizes, the NGDP Rule will continue to call for base growth in excess of 5 percent so long as the GDP gap remains negative.  That is, more QE should be in the works.

Should the Fed move to explicit nominal GDP targeting?

Proponents of nominal GDP targeting assert that it has several advantages over interest rate or inflation targeting.   First, compared to inflation, nominal GDP is relatively easy to measure.  There are numerous measures of inflation, but just one nominal GDP.  Second, it automatically addresses the Fed’s dual mandate of seeking price stability and full employment.  Third, it imposes an explicit anchor on monetary policy. 

A disadvantage is the need to adjust for changes in velocity and the need to pick the appropriate correction factor (value of k).  Also, because inflation is not explicitly targeted in the MR, opponents argue that inflation is more likely to be unleashed with an NGDP rule.  However, if nominal growth is credibly targeted at 5 percent, then inflation expectations are likely to remain subdued.  After all, with long-term real growth of 3 percent, nominal growth of 5 percent means long-term inflation is constrained to 2.0 percent. 

More QE is probably in our future.  By tying monetary policy to a nominal income target the Fed can at the same time provide a rationale for further monetary stimulus now and provide confidence that excess monetary stimulus will be extracted later.  Once velocity begins rising and nominal income reaches its target, under the rule monetary base growth will slow and eventually turn negative.  

¹There is more than one version of the McCallum Rule.  The one stated here targets the level of nominal GDP.  Other versions target the growth rate of nominal GDP.

I don’t think I’ve ever watched a compete 35 minute internet video.  I haven’t been to a movie for well over a decade and probably less than 10 times over the 40 years I’ve been married.  We have a TV at home, but no cable or antenna.  The kids use it to play Wii or watch videos.  I watch a video at home maybe once or twice a year these days.

Let’s just say, I’d rather read.

However, I just watched a full 35 minute video of a debate between Tyler Cowen and Eric Brynjolfsson.  Arnold Kling, another very good economist, calls the video “the video of the year,” which is why I watched it.  These are two really smart economist who are looking at the World and coming up with very different conclusions.  Here’s the video.

Tyler is the author of The Great Stagnation: How America Ate All The Low-Hanging Fruit of Modern History,Got Sick,and Will (Eventually) Feel Better, while Eric is a coauthor of Race Against The Machine: How the Digital Revolution is Accelerating Innovation, Driving Productivity, and Irreversibly Transforming Employment and the Economy.

The debate is fascinating.  The main question is if our current situation reflects stagnation or a “painful readjustment?”  Other questions pop up: Do we use median or average income?  Is technology causing unemployment, or a lack of technology causing unemployment?  Is wealth of total factor productivity the measure of progress?  What does GDP tell us about utility?

I recommend that you watch it.

A few days ago, I attended a presentation on by David Flaks. David is the Los Angeles Economic Development Corporation’s (LAEDC) Chief Operating Officer, and he was discussing LAEDC’s new LA County Strategic Plan.

A few hours later, I was called by a Ventura County Star reporter, Stephanie Hoops. She asked for my thoughts on the plan. I gave them to her, and she accurately reported them. They were on the web in a few hours and in my morning paper the next day.

For reasons that still elude me, some people have interpreted my comments as somehow dissing the LAEDC.

I had three comments:

• The plan was not very specific. Well, of course it wasn’t very specific. Los Angeles County is probably the World’s most diverse metropolitan area, with 88 cities, almost every ethnic or cultural group on earth, and dozens of languages spoken. The document was developed to attract broad concurrence. LAEDC went to a lot of trouble and effort to get everyone’s buy in. That would be impossible in Los Angeles County for any document with specificity. Any document that will attract broad support in Los Angeles County must necessarily be vague.
• The plan will probably not achieve its proponents’ goals. Here, I made the assumption that the goal was to create a measurable change in the Los Angeles County’s economic future. If that is the goal, then they probably won’t be successful. You just can’t do that with a vague document. Besides, many of Los Angeles’ challenges originate in Sacramento, and much of what goes on there is beyond Los Angeles’ control.  If, on the other hand, the proponents’ goals were less ambitious, say to begin moving Los Angeles County’s policies toward creating a more business-friendly environment, the plan could very well achieve its purpose. This brings me to my final comment.
• It was worthwhile to develop the plan. California faces serious and persistent economic challenges, and most of those challenges are a result of policy. We must move toward more business-friendly policies throughout California if we are to meet our obligation to provide economic opportunity to all Californians. This document, and many more like it, is a necessary step in moving California toward a consensus that encourages California policies that will result in the vigorous economy necessary to achieve the opportunity that Californians once took for granted.

I don’t see how this can be interpreted as negative toward the LAEDC. It is certainly not meant to be.

Finally, some people seem to think I speak for the University. That’s just not true. The University lets me say what I choose to say, but they don’t always like it. It is testament to the strength of the University’s commitment to academic freedom that it continues to support me even when I’m controversial.

Global population is estimated to reach 7 billion this year.  The United Nations is projecting that global population may reach 15 billion by the end of the century.  This projection is based on population growth continuing at its current growth rate of a shade under 1% per year.  However, extrapolation of the current growth rate is not a reasonable assumption.  Population growth is slowing dramatically almost everywhere, because of falling birthrates.  Demographers expect birthrates to continue to fall.  In all likelihood, global population will peak out sometime over the next 50 years well short of 15 billion.

The demographic transition describes demographics dynamics associated with rising wealth and improved medical care.  In agrarian societies before modern medicine, both birth and death rates were very high.  The first impacts of improved hygiene and medical care are declining infant mortality rates and increased life expectancy.  For both reasons, population growth takes off.  Then, with a lag, comes adjustment downward in the fertility rate.  Women have fewer children, partly because more children survive childhood and partly because fertility is negatively associated with wealth and education.

The process of falling fertility rates is happening everywhere.  In some countries, notably Russia, Japan and Italy, fertility rates have fallen well below replacement rates and population will soon begin to fall.  Even in those parts of the world where fertility rates are still pretty high, like Africa and parts of the Middle East, fertility rates are declining rapidly.  Naturally, it is difficult to say at what levels fertility rates will bottom out, but it is entirely possible that the equilibrium will be below replacement so that global population eventually peaks out and actually begins to fall.  Based on current trends, this could occur as early as 2050, with a world population of about 8-9 billion.

The scenario of declining global population would suggest very different problems from the UN’s 15 billion population scenario.  The problems of resource utilization and commodity shortages, which dominate the UN agenda, would be much smaller.  The problems of providing for an aging population would loom larger and so would government budgetary challenges.  The ratio of working people to non-working people would decline and growth in the standard of living (output/population) would slow relative to the growth in productivity (output/employment). 

If productivity growth is huge, an aging population may not be all that big an issue.  A relatively small working population may be able to produce sufficient goods and services to satisfy everyone’s demands.  This scenario could be achieved by combining a high savings rate with education and innovation, but there would still be problems.  This scenario implies a high tax burden on workers, who are likely to resist the high tax rates and the high savings rates necessary for the transition.

Economic models suggest that savings rates have a life cycle, low for young people and the elderly, and highest during mid-career.  This poses particular problems for the United States.  The U.S. savings rate, which has been quite low for decades, eventually will move even lower, pushed down by its aging population.  To offset this, it would be helpful if more people took on the responsibility of providing for their own retirement and ramp up their planned savings.

Will baby boomers (people born between 1947 and 1964) be able to save more, or is it already too late?  The oldest baby boomers are now reaching retirement age.  It is probably too late for them to build their savings, if they have not already done so.  The solution for them, in many cases, is to work longer.  On the other hand, the youngest baby boomers are still in their 40s and have plenty of time to build their asset portfolios.  Public policy can help in this regard.  For example, substitution of a consumption tax for the current income tax would support higher savings.

As people plan for the future a key assumption is the rate of return that can be expected on risky assets.  Many advisors suggest assuming that a balanced equity and bond portfolio will return approximately 8 percent, consistent with long-term historical trends.  But, historical returns may not be a reliable guide to future returns.  Future returns will be affected by trends in productivity and demographics, among other factors.  Higher savings rates are consistent with rising demands for risky assets and greater asset price appreciation.  Eventually, an aging population will be associated with lower savings rates as people sell assets to fund retirement spending.  This scenario applies today in Europe and will apply to the United States in coming decades.  I think it is prudent to plan for long-term returns somewhat less than historical experience.

To summarize, we will face enormous problems over the next 50 years or so, but they probably will not be the ones we currently expect.  The big problems won’t be overpopulation or running out of resources.  The problems will be dealing with aging populations, finding useful work for young and old people alike, and building retirement security in an era of low investment returns.

Today’s jobs report indicates our labor markets remain in the doldrums. The unemployment rate fell slightly from 9.2 percent in September to 9.1 percent in October, but jobs increased by only 80,000. The 80 thousand job gain was the result of private gains of 104 thousand that were offset by government sector losses of 24 thousand.

The October jobs increase was below than the revised September jobs increase of 158 thousand and below the July and August jobs increases.

Construction sector gains of 27 thousand in September were largely offset by 20 thousand in losses in October. The sector that gained the largest number of jobs was Professional and Business services with gains of 32,000 jobs. Other job-gaining sectors were Education and Healthcare, 28,000, Leisure and Hospitality, 22,000, and Retail, 18,000. Other sectors were little changed.

The long-term unemployed, i.e. those unemployed 27 weeks or more, fell from 6.2 million persons in September to 5.9 million persons in October. This is a seemingly welcome result, but it actually reflects more weakness in United States job markets. The decline is due far more to discouraged workers exiting the job market than to any underlying economic strength or job growth.

The broad measure of the unemployment rate, which includes persons marginally attached to the labor force and persons employed part time for economic reasons, fell from 16.5 percent in September to 16.2 percent in October.

Today’s jobs report indicates a labor market that still remains weak, with slow job creation and a high unemployment rate. I have argued in this blog-space that significant household sector debt reduction still needs to occur to get back to a healthy economy. Debt reduction has always been, and will always be, a long and painful process, particularly when a key asset price, housing, remains low.

The initial estimate of United States third quarter GDP was released today. The economy grew at 2.5 percent, driven mostly by consumption growth of 2.4 percent and investment in equipment & software of 17.4 percent. Growth was slightly augmented by investment in structures and the improvement in net exports. The government sector’s impact on GDP was zero, and the one detractor from growth was due to a fall in inventory stocking.

The jump in consumption growth from second quarter’s rate, which was 0.7 percent, implied that the savings rate fell. The BEA measure of the savings rate fell a full percent, from 5.1 percent to 4.1 percent. The economic growth estimate of 2.5 percent, with the approximate job growth estimate of about one percent, implies that output per worker growth was positive again in third quarter. It was negative in the second quarter.

What does all this mean?

Turning to the output per worker first, this measure of gross labor productivity, which contracted in quarters 1 and 2, increased in third quarter. Increasing labor productivity is a key driver of per-capita income growth, and it is a feature of increased innovation in production. It appears likely now that the negative labor productivity of quarters 1 and 2 were just temporary, and that productivity is returning to trend. If true, economic growth will benefit.

I am pleased to see the investment in structures and equipment/software. Investment raises the productive capacity of the future economy, thereby providing greater choice for either investing or consuming in the future.

The rise in consumption and fall in savings worries me. While it benefits current growth, it is likely to prolong the balance sheet rebuilding that I feel is necessary to ensure healthy growth in the future.

The question of which way the savings rate will go is also one of the big macroeconomic forecasting challenges of the day. Some Economists argue that because of the Great Recession, households will see that they need to rebuild their balance sheet, in this case mostly by reducing liabilities. Some Economists argue that the baby boomer generation, still very influential on the economy, is culturally incapable of saving. Both arguments have merits, and it is a tough call.

From a forecasting perspective, it is hard for the econometrician to see the factors that might drive the decision wether or not to save. The relevant factors are likely to include: household structure (married or not, kids or not), age, employment history, wealth level, debt level, type of debt, education level, and skill level.

Our forecast for third quarter GDP, 0.6 percent, was really low, and much of the error stems from our consumption/savings forecast. Our forecast presumed a similar savings rate to second quarter, and thus a very slow consumption growth rate. Given that it is difficult for me as a forecaster to see the above-mentioned savings-decision factors, I am not sure if my forecast was based on what was more likely, or if it was simply what I hoped.

Most indicators of household debt indicate to me that debt levels are still too high. As a result I hold to my belief that long-run United States economic growth will rise if households save now, i.e. pay off their liabilities. However, I am not sure if or when this will happen.

“Value at Risk is like an air bag that works well all the time except when you have an accident”

David Einhorn

Value at Risk (VaR) is a well-accepted measure of market risk.  It is defined as the minimum loss that could be expected to occur over a specified (short) horizon with a specified (low) probability.  For example, the 99% one day VaR would be the amount of daily loss that should be expected to be exceeded one day in a hundred.  The VaR measure is ubiquitous today; part of the risk measurement or risk management process of nearly all commercial banks, investment banks and insurance companies, and many if not most money management firms and hedge funds.  Regulated companies typically have minimum capital standards that are tied to a VaR calculation. 

A nice thing about one day VaR is that you get a lot of feedback on the quality of the measure.  Each day you compute your actual gain or loss and then compare that to the VaR calculated at the prior day’s market close.  If the calculation is reasonably accurate, you should expect to observe actual losses in excess of the calculated 99% VaR 2 to 3 times per year.  If you cannot produce a VaR calculation that meets this criterion then your risk management process needs some work.  By observing the trend in VaR over time, you can get a sense of the trend in the amount of risk that is being taken.  That is, unless the nature of the risk changes a lot.

But having a successful daily VaR calculation does not mean that you are prepared for an extreme scenario.  In an extreme scenario, you might easily see losses many times the VaR.   This is partly because the typical VaR calculation is based on recent data and most of the time recent data does not include extreme scenarios.  It is also partly because the VaR focuses on the minimum loss that could be expected every hundred days or so.  The VaR does not estimate how serious the loss could get in the extreme event.  Nor does it capture the effects of a series of successive VaR violations.  It is generally assumed that each day’s market move is independent of the prior day’s move.  In a severely negative scenario this is not likely to be the case.

Criticisms

As indicated by the above quote, VaR has come in for its share of criticism.  Some critics argue that VaR is worse than useless because it gives a false sense of security.  Others claim that it amplifies cycles.  Are these criticisms deserved?  I think the answer is that VaR is highly useful for its intended purpose as a risk measurement and monitoring device.  But, it is not the proper tool for estimating potential losses in an extreme scenario.    

Many people associate VaR with the worst case outcome.  This interpretation can lead to problems.  If the calculated VaR is small relative to capital, naïve senior executives might incorrectly conclude that the firm’s risk profile is too low, and that they should take on heightened risk. This is an understandable but unfortunate misreading of the meaning of VaR, which is that it is reliable only for normal times.

If you are trying to assess tail risk, VAR should be supplemented by a methodology that aims at estimating the magnitude of loss that could be expected to occur in an extreme event.  This is the province of stress testing.  Stress scenarios should be based on the worst case that we have seen or believe to be plausible.  It seems to me that capital levels for a company that intends to be a going concern should be based on such a calculation.  What happens in extreme scenarios for the major risk factors that impact your performance?  Does your profit/loss profile look like a smile or a frown?

My view is not shared by all.  Many commentators argue that hold capital against extreme scenarios is overkill.  Under this approach, they say, many useful investment projects would not be undertaken.

But, I think we have learned at least two things over the recent past.  First, worst case outcomes occur rather frequently, certainly more frequently than we expect.  Second, it is not good for long-term returns when investments are periodically wiped out.  I think it is reasonable for equity investors to expect that management maintains a sufficient capital cushion to handle extreme scenarios.