Relevant and even prescient commentary on news, politics and the economy.

income mobility

The Treasury recently came out with a study of income mobility in the US that is receiving a lot of play in the Wall Street Journal and in several blogs.

Income mobility studies are very difficult to do properly and often are very misleading because they incompletely adjust for different factors.

The biggest factor driving income mobility in the US is age. People start working as young people and as they gain experience and earn promotions they receive higher incomes. The Census Bureau reports data that shows how this process. According to their data in 2006 median incomes by age bracket were:




% ch















0VER 65



2006 is typical or normal. If you looked at the data in 1975, 1985 or 1995 or any year in-between you would find a very similar pattern of income growing sharply as people age and starting to fall after the age of 55 and falling very sharply after age 65 when most individuals retire.

The Treasury study looked at income tax returns of people over 25. In this way they avoided the problem of showing the Doctors daughter making $5,000 at age 20 with her summer job as a life guard at the country club and $50,000 at age 30 as a drug company representative.

Income mobility is commonly described as an escalator moving everyone on the escalator up over time. It is a good analogy, and the three factors drive this move. One is economic growth. If you take a snapshot of everyone on the escalator in 1995 and another snapshot in 2005 you can see how the average has changed and get a picture of how economic growth influence the location of the escalator.

The second factor is the aging of the population. This is what dominates this Treasury study. They took a sample of the population in 1995 and came back and looked at the same individuals in 2005 when everyone in the sample was ten years older. This approach deals with some of the problems in the first methodology where the comparison between 1995 and 2005 would look at different people. In the first methodology in 2005 the people in the 15-24 year old age bracket were not included in the 1995 snapshot and the 15-24 year olds are now in the 25-34 year bracket.

There is nothing wrong with this methodology and it deals with some of the problems in the first approach. It is a valid approach. But you have to be careful in evaluating what the results mean. For example, if you look at what happens to the income of the lowest quintile you see that their income grows very rapidly. Their incomes will grow rapidly because of three factors. One is overall economic growth that moves everyone higher. A second is age which will move everyone in the 25-34 age bracket into the 35-44 age bracket, etc (See below).. A third is what we normally think of as economic mobility as individuals through hard work, education and good fortune or luck improve themselves. But since this study is dominated by people aging the lowest income group of the 25-34 age bracket in 1995 and in the 35-44 age bracket in 2005. Given that it is normal for the average income of the 35-44 age bracket to be around 25% higher than the 25-34 age bracket the study finding that over half the individuals in the lowest quintile moved to another quintile is not surprising. What is surprising to me is that about half of the lowest quintile were still in the lowest quintile ten years later.
















over 65


The problem with the Wall Street Journal and several blogs analysis of this data is that they tend to credit all the change from 1995 to 2005 to overall economic growth and/ or
individual economic mobility. But this is misleading as the dominant factor driving the changes reported in the Treasury study is the aging of the population. But we do not have the data to age adjust the data to see how much of the reported improvement was due to this factor.

This is not to say that there is anything wrong with the Treasury study, It is a common shortcoming of many such studies. That is the reason why the best studies on income mobility are the new series of studies that look at lifetime earnings and compare lifetime earnings of one generation to the lifetime earnings of their parents generation.

But it does mean that many of the advances in average income reported and incomes of the lowest income quintiles emphasized in the Wall Street Journal discussion of the study are misleading. They are attributing the sharp growth in the income of the lowest income brackets to overall economic growth and income mobility when it is largely due to people moving into the next age bracket. But they conveniently fail to point this out.


I see the history rewriters are tying to develop a new meme about FDR and the depression. They are now trying to argue that FDR scared businessmen and capitalist so badly that they were unwilling to invest and this was why the recovery was so weak and the depression lasted so long.

As usual, they take a historic fact, that investment in the recovery was weak, and blame it on the New Deal policies and conclude that the New Deal had a negative impact. As the chart below shows, real nonresidential fixed investment in the depression was very weak when you compare it to post war cycles. Note, that in the chart I have rescaled the post-war cycles so you can easily compare the differences. The data is rescaled in two ways. One, the post-war data is quarterly data while the depression data is annual data. Two, the post-war data is on the right scale and the depression data is on the left scale and the scales are set so that the magnitude of the downturns appears similar even though the depression downturn was much larger.

The argument is that the weaker depression rebound was because FDR scared investors. It ignors the possibility that other factors that normally drive investment might have been responsible. Over the years as a business economist I have concluded that three factors dominate business fixed investment. They are corporate profits, capacity utilization, and the stock market. The stock market is earnings times the market PE. So using stocks as a variable captures the cost 0f capital as the market PE is it’s inverse. But it also means the variables double counts profits. For capacity utilization I use the gap between potential real GDP published by the Congressional Budget Office and actual real GDP. Not to get into a detailed discussion, but the following chart shows how well changes in the three variables have explained changes in real nonresidential fixed investment since WW II.

So the next step is to see how well these variables explain real nonresidential fixed investment during the depression. If they do a poor job it might imply that New Deal did scare business and capitalist into not investing. There is one problem in doing an exact comparison. The CBO does not calculate a real potential GDP series for the pre-war era so I can not calculate the GDP Gap for the earlier era in the same way. But it is an easy problem to surmount. From 1900 to around 1975 trend real gdp growth in the US was 3.5%. as this chart shows. So it is easy to substitute this trend real GDP series for potential real GDP to calculate the GDP gap.

So what did I find? I find that these three variables, profits, the GDP gap and the stock market do a great job of explaining real nonresidential fixed investment during the 1930s. One of the interesting points to make is that the equation does an outstanding job of capturing the turning points at the 1933 bottom and around the 1937-38 second recession.

In two years, 1935 and 1936 actual real capital spending was right at the bottom of the one standard error band and in one year, 1937, capital spending was moderately higher then the equation implied it should have been. But it has to be obvious that overall, the three factors of profits, capacity utilization and the stock market do an outstanding job of explaining business fixed investment in the depression just as they do in the post-WW II era. On balance this work clearly implies that the argument that FDR scared businessmen and capitalist into not investing is strictly a myth that is not at all supported by the data. Compared to post WW II cycles capital spending was weak in the late 1930s, but it was weak largely because of the massive excess capacity created by the economic collapse from 1929 to 1933, not anything FDR did.

No single approach to the dollar works all the time. But the most important is interest rate spreads.

The reason the dollar appreciated so much in the early 1980s is that interest rate spreads exploded upward to attract the foreign capital needed to finance the structural federal defictit created by the Reagan tax cuts. Int he early 1980’s “crowding out” worked through the dollar to crowd out the manufacturing and other sectors sensitive to the dollar rather then the interest rate sensitive sectors. Later after a new structure of large foreign capital inflows to the US economy been established as the new norm the currency moves did not have to be as large as they had been in the early 1980s. But interest rate spreads were still the dominant factor driving the exchange rate.

Average hourly earnings

Earlier this week Econlog and Marginal Revolution had a good discussion about an article by
Terry J. Fitzgerald in the Minneapolis Fed Review.

The article goes into a great deal of detail about the weak performance of average hourly earnings in recent years.

One point in the discussion was the new experimental measure of average hourly earnings being developed by the BLS. This measure has been getting a lot of attention lately with the widespread expectations that the new measure would show that the old measure of average hourly earnings has been significantly understating the rate of gain for average hourly earnings.

We now have some 16 months of results from the new experimental measure of average hourly earnings. it shows.


Mar-06 100 100
Apr-06 101.50 100.48
May-06 100.00 100.66
Jun-06 99.95 101.09
Jul-06 101.10 101.45
Aug-06 100.35 101.75
Sep-06 101.45 101.99
Oct-06 102.25 102.36
Nov-06 101.90 102.66
Dec-06 102.70 103.14
Jan-07 103.15 103.32
Feb-07 103.60 103.69
Mar-07 103.30 103.99
Apr-07 104.50 104.23
May-07 103.40 104.65
Jun-07 103.45 105.14
Jul-07 104.30 105.44

In the first year the new measure grew 3.3% while the old measure expanded 4.0% and over the first 16 months the old measure of average hourly earnings grew 5.4% as compared to a 4.3% increase for the new experimental measure.

So far it looks like the new experimental measure is demonstrating just the opposite results that what was expected by those who believe that average hourly earnings has been significantly understating the income growth of the about 80% of employment that punch a time clock rather than earn a salary or commission or are self employed.


There is a widespread belief that the destruction of industry in Japan and Europe played a major role in US economic growth in the 1950s. While the war time destruction did play a role in the lack of competition in the 1950s, the data says that it did not make a significant contribution to real GDP growth. If you look at the real trade contribution to real gdp growth you find:

1950-59 -1.70
1960-69 +0.02
1970-79 +1.31
1980-89 -1.21


Since the early 1980s the US economy has undergone a significant shift with the old 3-4 year business cycle giving way to much greater economic stability as since 1984 the US

has experienced only two minor recession. It appears that the US has shifted from having a recession roughly every 4 years to one every 10 years, or even less frequently. This was one of the justifications for higher stock market valuation and the great bull market of the late 1990s. But since 2000 the market PE has returned to the same relationship it had to interest rates and inflation that prevailed before 1995. Of course this implies that the stock market has gone back to discounting a 3-4 year economic-stock market cycle.

Generally, three or four main factors are credited with this change.

One is the improvement in information technology so that firms no longer make the inventory errors that generated recessions. One example of this was just completed. Last fall firms started to accumulate unwanted inventories as the inventory/shipments or sales ratio started rising. In the old days this would have lead to a major inventory accumulation that was followed by a sharp drop in output or a recession as firms finally liquidated the unwanted inventories. But this time firms reacted very quickly and liquidated the unwanted inventories before the problem became so severe it caused a recession rather then just a slowdown.

The past year’s inventory correction also demonstrated a second cause of the great moderation–the growing importance of international trade. In the old days if a retailers wanted to cut inventories it would cut back orders to a domestic manufacturing firm and that firm would have to cut output and lay off workers and this generated a negative Keynesian multiplier. Now, a great portion of the drop in orders is to China and other foreign producers who have to make the correction. We now export the recession and it shows up in the data as a drop in imports that actually boost reported real GDP growth. But this also means the US no longer gets the big snap back in output when the inventory liquidation is over. This was one reason the last two recoveries were weak by historic norms.

A third factor is a lack of shocks. Much of the poor economic performance of the 1970s was blamed on the oil shocks so the absence of such oil shocks was credited as part of the reason for the great moderation. But recently we have experienced another oil and commodity shock and it is not generating a recession as it did in the 1970s. Consequently, the luck or shock explanation for the great moderation is fading away.

Tightly tied into the impact of shocks or luck is the fourth factor explaining the great moderation is improved productivity. In the 1970s productivity was poor and unit labor cost rose so much that firms had little choice to pass through the higher oil and other commodity prices. But this cycle productivity has been so good that firms could absorb much of the higher oil and commodity prices and still sustain strong profits.

The final factor used to explain the great moderation is better policy, especially better monetary policy. As far as fiscal policy is concerned there was a significant policy shift in the early 1980s from tax cuts or deficit spending to stimulate demand to tax cuts or deficits to stimulate supply. Does demand create its own supply, or does supply create its own demand? For monetary policy it looks like the shift was from policy rules that gave inflation and unemployment roughly equal weight in determining policy to policy rules that gave fighting inflation a much greater weight.

So how do we evaluate these changes. Roughly, we have traded-off having frequent, severe recession for having much less frequent and milder recessions and lower inflation for about a 0.5 annual percentage point slowing of trend real GDP growth or about 0.25 percentage point weaker per capita real GDP growth So what does this look like. Maybe comparing actual real GDP to potential real GDP is one way to make the comparison. Potential real GDP is a measure of capacity calculated by the CBO based on growth in productivity and labor force. Thus, real GDP is a % of potential GDP is a measure of capacity utilization, or of actual economic performance relative to US economic capacity.

As the chart shows there was a major structural shift around 1980. Prior to 1980 actual real GDP was above potential real GDP almost two-thirds of the time while since 1980 it was only above potential about 2.5% of the time. It is like the old Army recruiting slogan, “Be All You Can Be”. Before 1980 it looks like the economy was “ All It Could Be” while since 1980 it has been “Less Than It Could Be”. Moreover, since 1980 as we experienced much more idle resources real potential GDP growth also slowed significantly as the following chart shows. The chart has two trend lines. A straight line for productivity pessimist and a curved line for productivity optimist.

When I look at these charts I reach a couple of conclusions. One, it looks like the difference in the pre-1980 era and the post-1980 is that the shift in monetary policy to giving inflation a greater weight than employment worked through the mechanism of creating sufficient excess capacity in the economy—a form of Phillip Curve analysis . Second, in contrast to supply-side policies Keynesian demand management worked to generate too strong an economy before 1980.. But since1980 supply-side policies have massively failed to stimulate strong economic growth. Of course, there are other explanations. One is that the strong economies of the 1950s and the 1960s was due to the stimulus and demands stemming from the Korean and Viet Nam wars. Of course that just leads to the conclusion that the Bush administration is just not really trying to win the Iraq war because it is giving the military the resources they need to win the war. Leave it to team Bush to provide virtually the only historic example of a was not stimulating he economy.

Otherwise, I’ll just leave these comments and charts as the start of a good discussion.



There was a discussion of Greenspan as a Fed Chairman earlier today.

Different people were making various claims with no evidence to support their positions.

So I though I would put in my two cents worth, but provide a chart to show why I take the position on Greenspan I do.

The first chart is my Fed Policy index. It is my version of the Taylor rule, but the biggest
difference between my version and other versions is that this one gives inflation and
unemployment equal weights. Most Taylor rule approaches gives inflation a weight of some 2 to 3 times economic slack. The others use the GDP Gap rather then the unemployment rate, but the difference is not significant. Moreover, the GDP Gap is a quarterly series iso t is not available as quickly. This chart uses smoothed monthly data and on an unsmoothed basis the index is now at 4.8%. Other Taylor rule approaches generally says rates should have been higher in the pre-Volcker era and about the same in the Volcker era then they actually were.

This version says that Greenspan started office by tightening more then was necessary as in the late 1980s actual funds were higher then the rule implies they should have been. This obviously contributed to the 1990 recession.

But the index implies that through the 1990s Fed Policy was almost exactly what the index called for. In other words we could have programmed a computer to give us almost exactly the same policy that Greenspan actually produced. Maybe the more interesting question is, if Greenspan followed the same policy rules as pre-Volker Chairmen had, why did it generate such different results. This clearly implies that K Harris is correct to think that much of what Greenspan achieved was strictly a matter of luck. Or maybe it was the lagged impact of the unusual tightly policies Volcker implemented. But again in the early 1990s actual policy deviated from the policy index as policy was significantly more expansionary then the index implied it should have been. We are still not certain if that was the best policy.

In this chart I simply did a regression of actual funds against the index and a dummy variable for Volcker. It is just another way of showing that through the 1990s Greenspan
did almost exactly what my policy rule called for.


Productivity growth is clearly slowing, and the key question is, is this a purely cyclical development or is it signaling a slowing in the long term trend growth rate.

Slowing productivity is a normal cyclical development and it is such a strong pattern, that historically, it has been a very good leading indicator. But this implies that we can not really determine if the recent slowing is just cyclical weakness or a trend shift.

To answer this question we need to look at indicators that either lead or determine productivity growth. It is a question economists have been researching for years

But clearly a major factor in productivity growth is the real capital stock per worker and it is generally widely accepted that this accounts for a quarter to half of productivity growth. Moreover, it also leads productivity growth by about a year.

There is also a long term relationship between trend productivity growth and trend growth in the real capital stock per workers as the following chart demonstrates.

From WW II until the late 1970s, early 1980s there was a very strong and relatively stable trend growth rate for both productivity and real capital stock per employee. But in the 1980s economic cycle growth in the real capital stock per employee stagnated, as it was still the same in 1990 as it had been in 1982. There was a cyclical bump in the early 1990s but this was driven more by a drop in employment then by stronger investment. With the capital spending boom of the late 1990s growth in productivity and the capital stock per employee surged again. But in recent years growth has stagnated again as since the Bush tax cut it has failed to grow.

This data on growth in the real capital stock per employee strongly implies that the recent slowing in productivity growth is not just a cyclical slowdown. Rather it looks like a shift to a slower growth trend driven by weak capital spending.

Obviously, given the limitations of blogging I can not get into much of a discussion of why growth in the real capital stock per employee did not grow in the 1980s or in recent years. However, these results are exactly the opposite of what the advocates of supply-side economics or republican trickle-down policies have promised. Maybe the simple answer is that there is no significant relationship between tax policy and capital spending.

But on the other hand since WW II the US has experienced good growth in the capital stock per employee except in the years when supply side economic policies were in place. I’ll leave it to the advocates of such policy to explain these discrepancies. However, I believe this is the fundamental or core reason why Cactus keeps finding that the economy does better under democrats than under republicans

Government VS private waste

In looking at the question of whether business or government is more wasteful it might actually be possible to use the pro-business argument to demonstrate that business is actually more wasteful.

The argument is that government does not have competition to force it to quit doing something so it continues to do things when they are no longer needed. I agree it is a valid argument.

The argument continues that competition forces inefficient firms to close down and they commonly cite examples of bad firms like Enron or Countrywide to document their case.

But they make their case with a few examples and virtually never look at aggregate data.

There was a study in the BLS monthly Labor Review of May, 2005 that actually looked at this issue.

The looked at how new businesses survive. They found that 33% of new business fail in the first two years and 56% fail within four years. I’ve heard these types of numbers cited for restaurants, but this study found that there really was no significant difference between the failure rate for restaurants and other industries.

The study also looked at other studies of the survival rate or exit rate for established firms. The studies they cited found that for established firms about 50% go out of business every four years and over 60% exit every five years.

The numbers are not exactly comparable, but the BLS study found that roughly 20% of US firms close down every year. That has to be a tremendous misallocation of resources. Remember, these will generally be smaller firms. But it is like the data on job creation. You hear all the time that small firms create most jobs This is true, but they also destroy more jobs so that over time their share of employment never seems to grow.

So what we are looking at is that government is wasteful because competition does not force it to quit doing some things. But on the other hand having some 20% of US businesses closing down every year also seems to be very wasteful. I’m deliberately using the term wasteful, not efficient because efficiency and wastefulness are not exactly the same thing.

On balance I do not how to make a meaningful comparison of these two phenomena. One reason is that I have no estimate of how much of government should be closed down.

So I have no idea whether the private sector or government is more wasteful. But I doubt that very many of the individuals making the case for private business not being wasteful are really aware how wasteful private business really is even though hand having 20% of US businesses closing down every year may be highly efficient because it is the best we can do.