Tuesday, August 31, 2010

Real GDP Growth in the US and Japan: A Closer Look at Consumption, Government Spending, Net Exports, Investment, and Inventories

In a previous post on the historical "stocks" of wealth in Japan and the US, I promised a post on the flows, i.e., aggregate income (GDP). I imagine someone must have posted versions of these before but for some reason I haven't come across them (just charts for recent quarters). Here are three longer-term charts followed by some comments.

Japan: Contributions to Percent Change in Real Gross Domestic Product (1981 - 2008)

(click on chart for a larger version in a new window)

US: Contributions to Percent Change in Real Gross Domestic Product (1980 - 2009)

(click on chart for a larger version in a new window)

US: Contributions to Percent Change in Real Gross Domestic Product (2005/Q1 - 2010/Q2) (SAAR)

(click on chart for a larger version in a new window)

Technical Note: The US data appears inconsistent if you compare the quarterly chart to the annual one. This is unmodified data directly from BEA's site so I have not introduced errors. It appears their annual data uses an average value of each flow (GDP, consumption, etc) for within each labeled year, rather than the end of year value. This would explain why, for example, the change in GDP for 2008 appears to be flat (i.e., 2008 average compared to 2007 average, rather than comparing end of year values) while the quarterly data confirms the plunge that was taking place in the second half of 2008.

Some Comparative Observations on Japan's Post-1990 Balance Sheet Recession and the US Post-2007 Balance Sheet Recession

Real GDP
Despite slowing in the early 1990s, real GDP in Japan only contracted in a single year — 1998 (at least, up until the 2008 global financial crisis). So far the US will only have experienced [just over] a year of actual contraction, though it was deeper. There is of course still the risk of "double dip", i.e., renewed contraction.

Household Consumption
Household consumption made a positive contribution to growth in Japan in every year except 1998, despite slowing after 1990 from over 2% to below 1%. In the US, consumption's contribution has been negative five out of six quarters starting in Q1 2008, but has since been positive in the quarterly data. Even if it remains positive in future quarters and years (TBD), it seems likely that ongoing deleveraging will cause the rate of increase to be smaller than the roughly 2% rate in the housing-bubble-and-home-ATM 2000s, which was in turn slower than the 1990s rate that ended at over 3% annually in the "new economy" tech bubble years.

Japan's household consumption was likely also supported by the fact that the household savings rate fell from around 13% in 1990 to around 3% in 2006. The US household savings rate is around 6% now, and unlikely to match Japan's decline (as it would have to go quite negative!). So it remains to be seen whether government deficits (boosting income) instead provide sustainable support for household consumption growth.

Direct Government Spending
Note that the measure shown is direct government spending, investment, and inventories, so it may be significantly less than the total government deficit during the same period. Hence these charts don't show the effect of the government giving the private sector more money to spend as it [the private sector] chooses, through lower taxes, larger transfer payments, etc.

Government spending growth represented about 0.5% to 1% of annual real GDP growth through most of the 1980s in Japan, and this increased a bit to over 1% in 1992 and 1993, likely in response to the deflating stock and real estate bubbles. (Were there large fiscal stimulus packages during these years?) Perhaps government spending going negative in 2007 is what helped tank the Japanese economy by 1998 — Richard Koo has mentioned pressure at that time to reduce government deficits and that this of course made the situation worse! The mid 2000s in Japan actually saw direct government spending contracting mildly each year.

In the US, government spending has grown every year except 1993, though at generally less than 0.5% of GDP it is not huge relative to other growth factors. Looking at the post-2007 recession and crisis response spending, the contribution is surprisingly small, at less than 1% in all but two quarters, and sometimes even negative! Part of the explanation is that this total includes state and local governments which have contracted to offset some of the federal spending, but the other possible partial explanation is that the Japanese may have been more aggressive in post-asset-bubble direct government stimulus spending (again ignoring tax cuts and other ways of government sustaining demand).

Private sector investment includes residential housing construction, commercial real estate, software, equipment, and other capital formation.

In Japan's late 1980s bubble, private investment contributed between 1.3 and 3.4 points to GDP growth each year. After slowing in 1991, it contracted for three years, taking almost 2 points off GDP in 1992 and again in 1993. 1998 saw a similar contraction.

In the US, investment was usually over 1% in the 1990s, but made a surprisingly small contribution to GDP in the mid 2000s housing bubble years (not much more than 1%, so much less than Japan's bubble-era investment). However, contracting private investment was the biggest contributor to falling GDP in late 2008 and early 2009. Late 2009 and the first half of 2010 have seen large a large bounce back in investment, perhaps reflecting the depth of the decline. Where might it go from here? I'll save that for a future post on outlook, but there is no obvious answer based on these charts alone.

Changes in private sector inventories are the primary driver of typical recessions and business cycles (if inventories are increased by too much as optimism gets ahead of reality, the recession allows the inventories to be drawn down, but reduces income to manufacturing and other contributions to inventory growth during the recessionary period). But in balance sheet recessions arising from debt-fueled asset price bubbles, inventory effects are dominated by other factors, particularly private investment. This is evident in the Japan chart from 1990-1993 and the US quarterly chart in 2008 and early 2009.

To compare to more typical recessions, look how much larger the green (inventories) is relative to the yellow (investment) in the US in 2001, 1982, and 1980.

There has been a huge inventory rebuild in the US in Q3 2009 through Q2 2010, so this effect will likely fade and be more neutral going forward (though with risk of renewed contraction in the immediate quarters ahead if the rebuild was too large or consumption falters further).

Net Exports
One common statement about Japan is that they are an "exporting nation" and that this sustained them after 1990 and prevented a depression. But this chart shows that in the 1990s, changes in net exports were sometimes positive and sometimes negative, and did not dominate the trend in GDP growth at all. However in the 2000s net exports have made a moderate and steady contribution to growth every year from 2002-2007.

Through most of the 1990s and 2000s in the US, changes in imports and exports exerted a mild drag on growth (i.e., imports rose faster than exports, as an increasing amount of income was spent on imported goods and services). This is a little surprising for the 2001-2005 period, as the dollar was falling most of that period. A deeper dive into the data could answer this, but on the surface it seems the growing demand for imports exceeded the increased competitiveness of exports. Perhaps China's currency peg was a dominant reason. Since 2007, growth in net exports has provided some positive support for US GDP growth.

This was a limited comparison narrowly focused on one type of data set (contributions to real GDP) and not tied into historical accounts of the timing of recessions, government stimulus, etc. I'd welcome a bit more elaboration in comments on the real world factors that contributed to the GDP growth trends for each country and time period (or any other insights into the data), if anyone is inspired.

As a future project, perhaps I'll eventually correlate these charts with changes in private sector debt and changes in government deficits.

Thursday, August 26, 2010

International CPI Trends: No Deflationary Spirals Evident So Far

While the US deflation risks seem to dominate discussion on the blogs I follow, there has been ongoing talk of the international deflation risks, especially for other housing bubble countries. The talk intensified in late 2009 as Ireland and Japan both experienced intensifying deflation. Discussion again picked up in spring to early summer of this year in the context of the anticipated deflationary pressures that austerity measures would bring to countries within the European Union.

Here's a long due update on price levels graphs for various housing bubble countries (the US, Australia, UK, Ireland, and Spain). Out of interest, I also include Japan, Iceland, Greece, and the EU27 area, as each has played a prominent part in crisis headlines over the last couple years.

Here are the actual price levels (not rate of change):

(click on chart for a larger version in a new window)

  • Price levels in Japan and Ireland appear to have bottomed out around January of this year.
  • The US suffered the biggest decline in 2008 and has yet to reach its previous price level peak.
  • The remaining countries have continued an upward price level trend, each to a different degree.
I had to stop including three-month rate-of-change graphs, as a lack of seasonally adjusted price level data for Europe made the result way too volatile to be useful. But here is the year-on-year inflation rate graph, i.e., the rate of change of the price levels above:

(click on chart for a larger version in a new window)

  • Inflation in the UK and Australia currently seems to be flattening out in the 3% range.
  • Inflation in Greece has kept rising and is over 5%.
  • Inflation in the US is falling.
  • Inflation in Spain and the EU27 region is around 2%, with the trend unclear.
  • Japan and Ireland both have a moderating level of deflation, moving toward flat line (this is more visible in the prior price level graph).
Will any countries within the EU experience a deflationary spiral as some commentators warn? It seems unlikely. Will they experience falling inflation and eventual possible mild deflation, following the Japan (and perhaps US) path? Maybe in some cases, but absent acute crisis conditions (a possibility), these price trend changes seem to take years to unfold. In addition, actual house prices don't seem to have corrected nearly as much in other countries as in the US (measured relative to incomes and rents). In the US house prices peaked around five years ago and while we have continued disinflation we don't yet have sustained deflation.

Here is the exact same graph zoomed out to a larger Y-axis scale to show Iceland in full:

(click on chart for a larger version in a new window)

Iceland is the extreme outlier of the group, with the inflation rate peaking over 20%, and falling continuously in the last year and a half. I don't know all the reasons for Iceland's experience, but currency changes must be a large factor. In 2008 its currency value crashed to less than half of its previous level against the dollar, as shown in this ISK to USD chart:

(click on chart for a larger version in a new window)

And as of 2008, it was more dependent on imports than most countries, with their value measured at around half of Iceland's GDP (via Google Public Data):

(click on chart for a larger version in a new window)

Thursday, August 19, 2010

Deflation Watch (July 2010): Still Disinflating Japanese Style, Actual CPI Deflation Probably Not Imminent

Three months of negative month-on-month CPI prints (a trend which broke in July) have generated an absolutely astounding amount of deflation commentary in the econoblogosphere! The Japanese-style-deflation-in-the-US meme has certainly reached critical mass in at least some circles...

I decided this month to finally update my charts. The format carried over from my original post comparing historical deflationary episodes overdoes the Great Depression comparison, despite being useful at the time (a year ago, negative year-on-year CPI reports had prompted a renewed surge in deflation chatter).

Some Relevant Current Articles

My list of relevant links has grown long, and in the interest of actually publishing a chart update today with the latest data, I'll save the links for a later post. (Some others are even labeling their posts "deflation watch" as well!)

Post-Bubble Consumer Price Index Trends: Current US (post-2007) versus Japan (post-1989) versus the US Great Depression (post-1929)
(click on chart for a larger version in a new window)

The above chart shows the actual price levels for the three commonly discussed post-asset-bubble deflationary episodes. Here are the year-on-year inflation rates for the same time periods:

(click on chart for a larger version in a new window)

Rather than attempting to align all three episodes of data at a CPI peak, they are instead now aligned at the dates that wealth (i.e., asset values such as real estate and equities) peaked. The goal is to compare the effects of the post-asset-bubble deflationary forces on the actual consumer price level trend in each case. I considered a few other possible alignment points, such as the peak in private sector borrowing (as debt-fueled asset bubbles tend to be the largest). However the peak private sector rate of debt-growth before the Great Depression was actually in 1925 (!), and overall the peak in balance sheet wealth seemed the most logical choice. The main reason for this choice is that falling asset values (along with a post-bubble consumer mentality) typically induce a higher attempted private sector savings rate, which will typically cause an aggregate demand deficiency and lead to downward pressures on prices and wages. Some charts of balance sheet wealth in the US and post-bubble Japan are in this previous post. The rough peaks I used (these are not terribly precise) for the charts were November 1929 for the Great Depression, mid 1989 for Japan, and mid 2007 for the US.

Key Observation: The US price level is currently following a similar path to Japan's experience (though slightly more rapid than Japan, still very far from the Great Depression experience), and it took many years of disinflation before Japan experienced sustained (mild) deflation. That said, the downside economic risks are much larger in the US today than in Japan post-1989. A number of factors still threaten a faster move through disinflation to deflation for the US today, including but not limited to:
  • Higher unemployment in the US today than 1990s Japan
  • Risk of negative global demand shocks due to such factors as deflating global housing bubbles (real estate prices in many countries are still nearer peak than trough by price/earnings and price/income measures).
  • Anti-deficit political pressure that threatens reductions in US fiscal policy spending
  • A low likelihood of any non-government sector (household or business) driving growth by reducing its savings rate significantly (Japan's consumers provided just this boost to demand by reducing their savings rate from around 15% in 1990 to under 5% by the early 2000s)
Still, it's probably premature to expect US deflation in the immediate months ahead.

Trends in Consumer and Producer Price Index Levels (CPI & PPI, Seasonally Adjusted)

(click on chart for a larger version in a new window)

The chart above shows various CPI and PPI measures, aligned at 100% on July 2008 when the CPI peaked. It is intended to help in comparing the trends in absolute price levels (not rate of change). For example, crude materials (PPI) and energy (CPI) are clearly more volatile than finished goods (PPI) or headline or core CPI, but you can see at times (especially late 2008) how they can drag the other indexes around.

Annualized 3-Month Rate of Change for Components of US Consumer Price Index, Seasonally Adjusted
(click on chart for a larger version in a new window)

This chart shows the short term (three month) rate of change of the components of CPI, to help identify underlying inflationary and deflationary forces within the overall CPI basket. Many components are quite volatile despite being seasonally adjusted.

A surprising number of components (food, housing, recreation, and overall headline) are showing a flat (0%) three month rate of change. But shelter (a subset of housing I have separated due to its importance) has broken its deflationary trend and continues with positive inflation! I am still looking for a good explanation (Calculated Risk has commented a little but is also unsure of details). If this continues, it removes a lot of the near-term deflationary impulse on the overall index, and perhaps actual lasting CPI deflation is still years away like Japan's was (or perhaps even not at all).

16% Trimmed CPI

The 16% trimmed mean CPI (generated from the Cleveland Fed site) removes the most extreme monthly price changes, and is still falling after only a brief interruption in the trend:

(click on chart for a larger version in a new window)