Tuesday, September 22, 2009

The Mystery of Japan's Private Debt Levels (Solved?)

For the last couple years there has been a lot of discussion of how the macroeconomic situation in the US compares to past historical episodes. Unfortunately the discussion of debt levels has relied mostly on anecdotal data, other than the one well covered comparison of the US today with the US at the time of the Great Depression (see for example Steve Keen's charts here). I'm among those who view post-1990 Japan as a highly relevant additional example. A few months ago I finally found what I hope is reliable debt data for Japan. I'll present charts of the data first but please read the sourcing methodology further down as it's possible this method of measuring of Japanese debt may not be an apples-to-apples comparison with the US Federal Reserve's debt data (please give feedback if you have it, and I will update the post).

Japan's Private Sector Debt From 1980 to 2007

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

The above chart shows the total debt in yen for Japan's household, corporate, financial, and government sectors from 1980 to 2007 (the latest available). Private sector debt actually kept rising (much more slowly) after 1990, peaking around 1997! Government debt also rose significantly during this time period. This large expansion of total debt likely contributed to the continued growth in GDP until 1997, seen here:

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

The next chart shows the debt-to-GDP ratios since 1980 — i.e., factoring in the impact of economic growth (as measured by GDP) reducing the debt burden relative to incomes. By this measure, total private debt-to-GDP did peak around 1990 when stock and real estate bubbles were bursting.

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

This suggests that in the context of a booming global export market and by continuing to increase its debt, Japan was able to keep growing for years after its asset bubbles burst. Interestingly the 1997-1998 time-frame is also when Japan's consumer price levels peaked (see chart here). Confirming anecdotal experience, the corporate sector was highly leveraged (much more so than the household sector), and did work on reducing debt but most notably so after 1997.

But the most striking insight this data yields is that Japan's economy from 1990 until 2005 did not deleverage on aggregate, due to the government increasing debt faster than the private sector was reducing debt, and that the private sector has only reduced debt from 386% of GDP (1990) to 328% of GDP (2007)! This government borrowing most likely explains why Japan did not experience a depression (and is somewhat consistent with the views of Richard Koo, who I have discussed before). But it reinforces the question of what the end-game is for Japan and whether it will ever be able to grow out of or pay down its massive total debt.

Note that I have seen and used in the past a couple other charts of Japan's debt-to-GDP ratios that disagreed with each other and with this data, for example this one: [UPDATE Jan 18 2010: I now believe the below chart is WRONG, see the end of this post]

I played with the Japanese Cabinet Office data I obtained for this post by excluding loan-based debt, excluding non-loan debt, excluding financial sector debt, etc and no matter what combination I tried I was unable to generate a chart that looked like this one. So one of the charts is probably wrong and it could be mine.

This article by Andy Xie, What We Can Learn as Japan's Economy Sinks, is very worthwhile reading on Japan's crisis and more (though there are a few things I'm not sure I agree with). The debt levels he cites for Japan differ a bit from what I derived, for example, he says "total indebtedness of Japan's non-financial sector is 443 percent -- probably the highest in the world, and far higher than the 240 percent in the United States."

Comparison with the US debt-to-GDP ratios around 2007

The following chart shows the rise in US debt-to-GDP levels by sector that many people have seen before:

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

It is color coded to match the colors of the Japanese debt chart's sectors for easier comparison. In the US, households carry substantially higher debt than they did in Japan. Here is a comparison of the debt-to-GDP ratios by sector at the start of each crisis (Japan around 1990, the US around 2007, plus also the US around 1929 at the onset of the Great Depression):

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

Household debt was much higher in the US in 2007 than past episodes, but Japan had much higher corporate debt and somewhat higher financial sector debt. Government debt was fairly comparable. The chart's debt-to-GDP numbers are in this table.

US 2007US 1929Japan 1990
Private Sector Total288%157%386%

Total Excluding Financials240%163%278%

In one respect this is a mildly optimistic outcome in relative terms — contrary to past claims (by me included), Japan's debt crisis is not dwarfed by the current US debt crisis. Depending on the US's political willingness to provide ongoing fiscal stimulus as Japan did, my opinion is this makes the Japanese style stagnation and mild deflation relatively more likely in the US, as opposed to a deep depression — though employment and GDP have already fallen more than they ever did in Japan (at least until this global crisis hit). Of course, total debt-to-GDP is not the only macroeconomic determinant that matters, so other factors could drive the US today to a different outcome.
  • Substantially higher household debt, especially with non-recourse mortgages standard in the US, could be a much more negative factor relative to Japan. There are far more distressed household balance sheets than corporate ones given the relative number of entities in each sector, so this could increase the system's susceptibility to full-blown debt deflation, especially given that households have less incentive to "extend and pretend" by faking solvency through relaxed accounting rules than corporations do.
  • The global context today is that most nations in the world have been involved in this crisis, many of them with high debt levels of their own. This probably explains in part why the current crisis has been deeper than Japan's post-1990, but the question remains whether this crisis is winding down or whether deflation will intensify and cause even deeper economic pain in the years ahead. With deleveraging barely having begun, there is strong reason to believe that we have years of adjustment still ahead.
  • Other differences in financial markets today such as the huge global derivatives market and associated counterparty risks could be relevant.
Many people summarize the options for removing excessive debt as inflate or default. While the long term result is uncertain, Japan has been carving a third option for nearly the last two decades — reduce nominal interest rates to very low levels to reduce the debt servicing burden while leaving the debt in place in an attempt to grow out of it, even if doing so is accompanied by economic stagnation. Note that this approach does not seem to cure insolvency, except perhaps through the long and slow process of using earnings for balance sheet repair. No doubt demographics are different in Japan compared to other countries and this has some impact, but the US and Europe are also increasingly facing aging populations.

Methodology for Obtaining Japan's Debt Data

I derived the data from selected balance sheet liabilities within the aggregate national accounts stock data provided by the "Economic and Social Research Institute (ESRI), Cabinet Office, Government of Japan." See "Part 2 Stock" under the heading "Accounts classified by Institutional Sectors". The super-set of liabilities listed across households, non-financial corporations, financial corporations, and general government are (the bolded ones are what I used):
  • Currency and deposits
  • Loans
  • Securities other than shares
  • Shares and other equities
  • --> Of which shares
  • Financial derivatives
  • Insurance and pension reserves
  • Other liabilities
Loans are clearly one kind of debt. It seems likely that securities other than shares are a good proxy for all other debt (e.g., bonds). Is this incorrect? This OECD definition of "securities other than shares" is:
"Securities other than shares consist of bills, bonds, certificates of deposit, commercial paper, debentures, and similar instruments normally traded in the financial markets."
The Federal Reserve Flow of Funds guide defines debt as follows:
"Credit market borrowing or lending is defined here as the transfer of funds through certain financial instruments: open market paper, Treasury and agency securities, municipal securities, corporate and foreign bonds, bank loans not elsewhere classified, other loans and advances (such as loans made under various federal programs), mortgages, and consumer credit. Excluded from the definition are a number of other items that are also sources and uses of funds for the sectors — official reserves, special drawing rights certificates, Treasury currency, deposits and interbank items, security repurchase agreements, corporate equities, mutual fund and money market mutual fund shares, trade credit, security credit, life insurance and pension fund reserves, business taxes payable, investment in bank personal trusts, proprietors’ equity in noncorporate business, and miscellaneous items; a sector’s credit market borrowing is thus not the same as the increase in its total liabilities."
It seems quite possible, especially for the financial sector, that the "securities other than shares" contain types of debt that are excluded from the US debt measures (for example, perhaps certificates of deposit). PLEASE TELL ME IF YOU KNOW WHETHER THESE MEASURES ARE COMPARABLE.

I did compare this data to a few individual data points I was able to find on Japan's sector-specific debt levels around 1990, and this data appears to be in line with those data points. The main unconfirmed data is Japan's financial sector for which I have found no other data points to compare.

Another Pre-Crisis Comparison: Sweden

The Swedish government's response to their early 1990s crisis has often been held up as a model for what the US should do, so it would be valuable to have an idea of the relative size of their crisis as well. As I summarized in an earlier post, the data points I've found for Sweden around 1990 so far are:
  • 170% total debt-to-GDP
  • 127% private sector debt-to-GDP (source)
  • 43% government debt-to-GDP (source)
It's not clear whether the private sector measure includes the financial sector, but if it does, Sweden's debt levels were much smaller than Japan's in 1990 or the US in 2007. However, they were more comparable to the US at the start of the Great Depression. Yet like Japan, Sweden was able to grow exports to a booming global economy in the 1990s. So while the absolute debt-to-GDP measure is central, I don't believe it is the only important macroeconomic determinant.

UPDATE 10/14/2009: I added a foreign debt row to the table of debt-to-GDP numbers (a commenter pointed out the numbers do not sum otherwise, so I probably shouldn't have skipped it in the first place).

UPDATE 10/14/2009: After originally posting, I did seek input on the derivation of this Japan debt data from a highly financially-savvy blogger, and he replied that "yes" it was comparable to the US data (though I can't guarantee he looked at it in detail).

UPDATE 1/18/2010: McKinsey has released a study on deleveraging that seems to validate my findings here. The only difference is that their numbers for Japan's financial sector are consistently around 50% of GDP less than what I found (perhaps reflecting exclusion of some specific bank liability such as certificates of deposit?). But the overall shape of the trends do match, most notably that public debt expansion has exceeded private debt reduction. Here is one of their charts:

Monday, September 21, 2009

Deflation Watch (August): Price Level Trends Relative to Past Debt Crises

This post updates the charts from Price Deflation Today versus the Great Depression and Post-1990 Japan — Comparative Charts with the data for August, and adds a series of charts depicting factors contributing to current deflationary pressures. Commentary will be minimal unless there is something noteworthy — the prior post includes lengthier observations on US Great Depression and Japan post-1990 trends that may still be relevant. My current plan is to update this post each month after CPI data is released as long as the comparisons remain interesting. (Unfortunately while I wrote most of this post last week I didn't have time to finish it then.)

As noted previously, "deflation" is often discussed in broader terms than simply price level:
  • Contraction of money and credit (broad money supply)
  • Deflation in asset prices
  • Deflation in a representative "basket" of consumer and producer prices
  • Deflation in wages
The various measures are often somewhat correlated but they only track to each other loosely. In the Great Depression prices fell faster than wages, yet wages (along with asset prices) still fell enough to propagate the adverse feedback loop of debt deflation in which income falls but debt obligations remain at the same nominal level, increasing the burden of the debt. Deflation in asset prices (triggered by the bursting of debt-financed asset bubbles) generally precedes the other deflationary trends.

CPI-U 12 Month Changes, 1999 to Present (US) (source)

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

BLS Summary Comments:
"On a seasonally adjusted basis, the Consumer Price Index for all Urban Consumers (CPI-U) rose 0.4 percent in August, the Bureau of Labor Statistics reported today. The index has decreased 1.5 percent over the last 12 months on a not seasonally adjusted basis.

The 0.4 percent seasonally adjusted increase in the CPI-U was driven by a 9.1 percent rise in the gasoline index. This increase accounted for almost the entire advance in the energy index and over 80 percent of the overall increase. Despite the August increase, the gasoline index has fallen 30.0 percent over the last 12 months.

The indexes for food and for all items less food and energy both posted slight increases in August. The food index rose 0.1 percent following a 0.3 percent decline in July. The food at home index, which fell 0.5 percent in July, was unchanged in August. Of the six major grocery store food group indexes, three rose in August and three declined. The index for all items less food and energy also rose 0.1 percent in August, the second consecutive such increase. Increases in the indexes for used cars and trucks, medical care, public transportation and lodging away from home offset a decline in the new vehicle index. The index for all items less food and energy increased 1.4 percent over the last 12 months, the smallest 12-month increase in the index since February 2004."

Consumer Price Index Trends: Great Depression versus Today through August 2009 (US)
(click on chart for a larger version in a new window)

Components of US Consumer Price Index (May 1927 - Dec 1937)

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

Components of US Consumer Price Index (Jan 2006 - Aug 2009)
(click on chart for a larger version in a new window)

Calculated Risk noted for this report the strange trend in BLS rent measurements:
"The BLS measure for rent increased slightly (rounded to flat). And owners' equivalent rent (OER), the largest component of CPI, increased slightly even though rents have been falling in most areas."

Price Index Changes: Great Depression CPI versus Current PPI through August 2009 (US)
(click on chart for a larger version in a new window)

Consumer Price Index Trends: 1990s Japan versus US Today (through Aug 2009) and US Great Depression
(click on chart for a larger version in a new window)

CPI in Japan (Jan 1980 - Jul 2009)

The peak of Japan's CPI occurred in October 1998, almost eight years after the stock market peaked, and Japan's notorious mild deflation has been in effect since then:

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

Factors Contributing to Deflation

This is a partial selection of measurable forces that contribute toward price deflation, though the core cause is the bursting of debt-financed asset price bubbles. Not all of these measures are in an outright deflationary trend, but most are suggesting at least disinflation. This data is US-specific, but US price levels will also be affected by global trends in the months and years ahead, for example with respect to whether China and other emerging economies falter meaningfully in their growth.

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

Inventory to Sales Ratio
(click on chart for a larger version in a new window)

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

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

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

Total Borrowing (All Debt Markets)
(click on chart for a larger version in a new window)

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

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

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


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

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

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

Household Wealth
(click on chart for a larger version in a new window)
Source: Calculated Risk

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

Thursday, September 17, 2009

Total Borrowing Continues Contracting in Q2

The Q2 2009 Federal Reserve Flow of Funds (Z.1) report shows a continuation of the Q1 trend — total borrowing in all US debt markets is contracting at an annualized rate equivalent to 2.1% of GDP. Government borrowing is falling slightly short of offsetting private sector deleveraging in Q2:
  • Government adding debt at a $2082 billion annualized rate
  • Private sector reducing debt at a $2515 billion annualized rate (household sector: -$233 billion, corporate sector: -$203 billion, financial sector: -$2079 billion)
  • Foreign sector adding debt at a $192 billion annualized rate
As I've stated before, I think this is the number one reason (among other reasons) why the huge supply of government debt is able to find buyers. Treasuries are simply filling a "hole" left by disappearing private sector debt! (i.e., on aggregate, previous holders of those debt assets will look for replacements, and treasuries are what is available.)

US Total Borrowing by Sector (Quarterly 2003 - 2009/Q2) Relative to GDP

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

The rate of private sector deleveraging (yellow line) is still increasing as of Q2. The rate of government borrowing (red line) has been mostly flat since Q3 2008, with a dip in Q1 2009. If these two trends continue, total borrowing may begin to contract more significantly, with potentially more noticeable macroeconomic impacts (for example, deflation).

US Change in Debt By Sector (Quarterly 2003 - 2009/Q2)

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

This chart shows the quarterly change in debt for each sector, i.e., borrowing relative to existing debt for that sector rather than relative to GDP. Since government debt is much smaller than private debt, it clearly is growing at a much larger rate in these terms. Also of note is that the private sector is deleveraging at a 6.6% annualized rate, which leaves a lot of room for further private sector debt reduction.

US Total Borrowing By Sector (1974 - 2009/Q2) Relative to GDP

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

This chart shows the same data on a longer time-line — annual data since 1974.

US Borrowing by Sector Excluding Bank Credit (1974 - 2009/Q2) Relative to GDP

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

This chart again shows the longer term borrowing trend but with changes in bank credit (TOTBKCR) subtracted from total borrowing and private sector borrowing. In other words, it captures the borrowing enabled by non-bank lending (such as bond issuance). While the data isn't very useful for most of the years shown (since expanding bank lending increases the money supply and amplifies the non-bank lending that can occur), the point of this chart is to show that the recent contraction in total borrowing is not about contraction in bank loans alone.

UPDATE: I have focused on total debt including the financial sector in part as context for addressing the eternally asked question of why government debt finds enough buyers, however total non-financial borrowing may have more relevance to the health of the economy due to a higher correlation (maybe?) with consumption markets rather than asset markets. By this measure, we are not deleveraging in aggregate because of current government spending, as the Z1 report makes clear:
"Debt of the domestic nonfinancial sectors is estimated to have expanded at a seasonally adjusted annual rate of 5 percent in the second quarter of 2009, about ¾ percentage point faster than in the previous quarter. Private debt contracted in the second quarter while government debt expanded."

Tuesday, September 8, 2009

Dividends, Earnings, and Stock Price Trends have Tracked the Great Depression

Most people have seen Doug Short's charts comparing stock price trends during this bear market with past episodes. However, I haven't seen the same comparison done for dividends and earnings and was curious to see the result. I put the following charts together primarily with Robert Shiller's stock market data, though I added in the operating earnings data for recent years from the official Standard & Poors S&P500 earnings spreadsheet. Stock prices use only a single data point for each month so these charts may differ a little from those that use daily stock price data.

Stock Price, Earnings, and Dividend Trends During the Great Depression

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

Leading up to the Great Depression, the stock market's peak value was in September 1929. The value of stock prices, earnings (12 month trailing), and dividends depicted in this graph are all relative to their value in September 1929, so the relative decline of each from that date is visible (earnings and dividends rose slightly after September 1929 before declining). My observations:
  • It is remarkable how closely the decline in prices tracks to the decline in earnings! However, the chart is a little visually deceptive since the lower the values are on the chart, the larger the percentage difference between the lines at that point, e.g., the gap in blue (price) and yellow (earnings) lines in mid 1932 is larger on a relative basis than gaps higher on the chart.
  • Also of note is how the decline in dividends lags the decline in earnings, with dividends ultimately bottoming out 55% below their peak value versus a larger 75% fall in earnings.

Stock Price, Earnings, and Dividend Trends in Recent Years (US)

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

The recent stock market peak was in October 2007. As with the previous graph, stock prices, earnings (12 month trailing), and dividend values are relative to their value in October 2007. My observations:
  • Just like during the Great Depression, it is remarkable how closely stock prices track to earnings! However, they have tracked operating earnings and seemingly completely ignored reported earnings. It appears that the push by Wall Street to focus on operating earnings (which exclude many supposedly "one time" charges, write-downs, etc) over reported earnings has been quite successful with respect to investor behavior. The gap between operating and reported earnings has grown significantly over the years so operating earnings are controversial.
  • Reduction in dividends again lags reductions in earnings.
  • As of Q2 2009, stock prices have jumped above the earnings trend "track" by the largest amount yet. Interestingly this occurred when reported earnings rose slightly even while operating earnings declined, so perhaps the market simply chooses the most optimistic of the two! Of course the changes this spring in accounting rules (with less mark to market) may have made quite a difference in earnings in 2009 and I'm not sure whether anything comparable happened during the Great Depression.

Stock Price, Earnings, and Dividend Trends Today versus the Great Depression (US)

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

This chart combines the previous two onto a single overlapping time-line, with all recent values relative to October 2007 values and all Great Depression values relative to September 1929. The lines are color coded to reduce confusion a little — red for dividends, green for earnings, and blue for stock prices, with the darkest of each color representing the data during the Great Depression. My observations:
  • While Doug Short's charts have already shown the similar stock price trend now compared to the Great Depression, the recent trend in dividends and earnings is amazing similar to the Great Depression trend as well! (At least if one compares operating earnings today with reported earnings back then, since operating earnings did not yet exist).
  • While anything could happen in future months, the current trajectory of dividends (down 11% from peak) is looking slightly steeper than the Great Depression decline. Since reported earnings have fallen so severely even relative to the Great Depression, and reported earnings are a closer representation of actual cash flow than operating earnings, the downward pressure on dividends seems like it should be much greater than during the Great Depression — unless reported earnings can stage a miraculous recovery. Is there a reason I'm not aware of why this conclusion is wrong? If correct, bonds may look increasingly attractive relative to stocks for income-seeking investors.

Stock Price/Earnings Ratio Trend during the Great Depression versus Today (US)

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

Stock Dividend Yield Trend during the Great Depression versus Today (US)

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

Closing Thoughts

To me, the clearest value in these charts is yet one more piece of evidence that this recession/depression is very comparable to the Great Depression, despite the claims of many people that this period is "nothing like the Great Depression." And I did check the data for the years since WWII and found no other dividend declines near as large as today's, so this is not a normal recession pattern. That said, these charts say nothing about the future and what will result from current government stimulus and any structural economic differences that may be relevant. Stock prices, earnings, and dividends could immediately start heading sustainably higher and break from the Great Depression trend, or they could keep heading down. My personal view is that the downside probabilities are much higher than the upside ones (with flat being another feasible path somewhere in between), but everyone needs to make their own determinations and none of this is investment advice.

Friday, September 4, 2009

Price Deflation Today versus the Great Depression and Post-1990 Japan — Comparative Charts

In recent months I've noticed an increasing amount of commentary on the intensifying deflationary forces in the US and global economy, for example:

Albert Edwards: "A Full Blown Deflationary Episode" Coming
"With import prices down some 19% yoy and even a record 7.3% yoy if one excludes petroleum, no wonder the price of domestic purchases has already fallen into deflation. If anything, domestic purchases inflation leads trends in both GDP and core CPI, so this is significant news."
Incidentally, Edwards also offers some confirmation of my expectation that banks will increasingly buy treasuries:
"But what about massive supply of government bonds I hear you ask? Won't that drive yields higher? Well it never did in Japan. But let's cast our minds back to the early 1990's US credit crunch (which seems so minor in retrospect!). What happened then is that US commercial banks bought US Treasuries aggressively at the same time as they contracted lending to the private sector (see chart below). This continued well after the end of recession in early 1991."
David Rosenberg has been outlining some similar deflation themes in recent newsletters as well.

Chris Whalen of Institutional Risk Analytics gets colorful in his latest newsletter, Q2 2009 Bank Stress Test Results: The Zombie Dance Party Rocks On:
"Despite all of the talk and expenditure in Washington, the US banking industry is still sinking steadily and neither the Obama Administration nor the Federal Reserve seem to have any more bullets to fire at the deflation monster."
So what exactly does this deflation threat look like and how does it compare to historical episodes of deflation? I decided to take a closer look at some of the data myself, and I'll share some charts below.

First, a clarification of what is meant by "deflation". While most people think of it as a reduction in [consumer] prices, others, for example Austrian economists, define it as a "contraction of money and credit", and this is a valuable viewpoint in part because broad money supply changes often precede changes in consumer prices. Both total private sector borrowing (see chart here) and bank credit (see chart here) have been contracting so far in 2009. Then there is of course asset price deflation, and there is no question this has been significant since the crisis began, for both real estate and stock market prices, and that this is one driver of reducing spending. This post focuses on consumer price inflation/deflation, even though the other measures are probably more important in general to understanding the macroeconomic trend.

To visualize Consumer Price Inflation, I've only seen graphs of period-based changes in the price level (month-on-month or year-on-year), such as this chart from the July 2009 BLS CPI report:

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

This is sufficient in normal times during which price levels only move in one direction (up). But with some price measures (especially energy) recently bouncing between positive and negative directional changes, I think the period-based change charts can be too volatile to use alone, since they don't show the big picture of where prices were in the prior period, despite being an excellent tool for focusing in on rate of change. Since I couldn't find any graphs of price levels (as opposed to change in price level), even though they must be out there, I made my own. Here is the first showing consumer price index trends in the US today compared to the Great Depression:

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

The place where all the lines meet on the chart (at 31 months) is normalized to 100% with all price levels relative to that peak. This point corresponds to July 2008 for the modern data and November 1929 for the Great Depression data. I have chosen to show several years preceding this point to give context to how quickly each measure had risen before peaking. Some observations:
  • CPI has been measured with different methodologies at different times, so this is not an apples-to-apples comparison.
  • After peaking in July 2008, headline CPI plunged at a faster rate than the onset of the Great Depression. No wonder (among other reasons) treasuries rallied so dramatically at the end of 2008 and then reversed as headline CPI started heading higher again!
  • Core CPI (excluding food and energy prices) has kept rising, but the pace of increase is slowing in 2009, even with energy prices relatively stabilized, despite coinciding with the largest monetary and fiscal stimulus in history! Is the battle with the "deflation monster" being slowly lost?
  • Looking at the BLS year-on-year CPI chart further above, the trough in the rate of change of core CPI of around 1% in early 2004 was almost four years after the dot-com stock market bubble popped, and over two years after the recession lasting officially from March 2001 to November 2001 ended! So (at the risk of being wrong by using only one data point) impacts to core CPI seem to lag recessions and changes in asset prices by years. It seems likely that this trend will replay today.
  • Headline CPI has turned (mildly) negative again in July — will a full downward trend resume? Might we face price deflation as ferocious as in the Great Depression? Personally I think it unlikely, but can't rule it out entirely.
  • I added energy prices to the graph because they have been so volatile (whether due to global demand exceeding supplies, rampant financial speculation in energy as an asset class, or some combination of both). One way of looking at the effect on headline CPI is that the violent energy price spike simply forced headline CPI higher between mid 2007 and mid 2008 and that this is the primary cause of current deflation. While partially valid, I don't think that's the whole story.
Next here are the components of the CPI to show how each is contributing now versus during the Great Depression. Here is a graph of all the CPI categories available in the Bureau of Labor Statistics data for the main Great Depression years:

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

I was unable to find information on the relative weighting of the CPI components from this era. My observations:
  • Leading up the the minor peak of CPI in November 1929, all price measures except food were already in a couple years of moderate deflation. Could that have made them susceptible to rapidly tipping into a self-reinforcing adverse feedback loop of severe deflation when the economy was hit by the deflationary shock of contracting credit and falling asset prices?
  • As prices declined, food fell the most dramatically (almost down to half its peak price), dragging down the overall CPI. Is this analogous to commodity prices today? It's my guess (without any specific knowledge) that food prices back then tracked more closely to pure agricultural commodities, if there was less processed food, restaurant food, etc available.
  • Electricity prices did not deviate from their mild downward trajectory until overall CPI bottomed after several years — perhaps they were effected by regulation, high labor costs, or some other factor?
Now here is a chart of categories of the US Consumer Price Index in recent years:

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

The thicker lines represent heavier weights within the overall index. Observations:
  • Most of the components with smaller weights (medical care, recreation, education and communication, other goods and services) have still been trending higher.
  • Apparel has been remarkably flat for years (recently trending up) — I would have guessed we'd see more deflation in these prices.
  • Housing is the largest category (weighted 43% of CPI) and it appears to be in a gradually accelerating downward trend. I (and others) expect this to intensify in the years ahead, thus dragging down overall CPI due to the heavy weight. Rents and owners' equivalent rents are heavily weighted, and they are starting to decline in the real world due to the huge excess supply of housing built over recent years with more and more being added to the rental market, as well as recession-specific factors such as more people sharing housing or apartments to save money rather than living alone. I have read that this CPI component lags real world changes in rent.
  • Food is the second largest category (weighted 16% of CPI) and it also appears to have started a mild downward trend. If this continues, it will also pressure the overall CPI downward.
  • Transportation is the third largest category (weighted 15% of CPI) and it already fell precipitously and has been gradually rebounding. maintaining upward pressure on CPI. Will it keep rising, level out, or start falling again? Any change from the current upward direction could have an out-sized impact on CPI, given that this has been the largest weighted component in a rising trend in 2009.
Comparing the mix of categories that comprised CPI in the Great Depression versus today, it seems likely that services have a much larger impact on consumer price level today compared to the Great Depression. I would guess that the higher the percentage of labor (versus commodity prices) in the cost of what we buy, the stickier prices will be. Commodities are repriced on a daily basis in the open market so prices can adjust fast. Most labor is not repriced nearly so often (since businesses don't bid for replacement employees every day!), though of course there has already been a trend of wage cuts and furloughs in this recession/depression. Is the basket of what we "typically" buy today more resilient to rapid price changes and thus a source of downside protection against a repeat of the Great Depression's severe deflation? I don't know but I bet someone has studied this...

Though it may be even less of an apples-to-apples comparison than comparing CPI from two eras, I decided to compare PPI (Producer Price Index) today to CPI during the Great Depression (PPI data is not available from then):

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

  • Producer prices have fallen at a far more rapid pace since peaking in July 2008 than consumer prices during the onset of the Great Depression. They have stabilized but not resumed a convincing uptrend.
  • With such a dramatic divergence between crude, intermediate, and finished goods measures, clearly the enormous upward spike in crude goods prices prior to July 2008 has made it easier for there to be a rapid decline following the peak, so it may not be worth putting much weight on this graph, except as food for thought.
  • However, some observers suggest that PPI is a predictor of future CPI.
  • While I won't go so far as to predict it, it would not surprise me if the China growth story (which has numerous skeptics with convincing arguments) falters within the next year or two, leading to a fresh crash in commodity prices. Could this tip PPI and later CPI into a downward spiral?
Another logical comparison would be with what happened in Japan after its stock and real estate bubbles peaked around 1990. Here is the earlier graph (of the US today and during the Great Depression) with Japan's CPI data added:

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

I used January 1990 (the peak of Japan's stock market bubble) as the anchoring point for Japan's data because there was no CPI peak at the start of their 1990s "balance sheet recession" (note I made a mistake graphing and the Japan data is shifted 6 months to the right in the chart from the anchor point.). I am surprised at how long it took the inflation rate to decline — it took 3-4 years for price levels to begin to flatten out. Even into the late 1990s, there was still no real price deflation, despite a huge deflation in asset prices. Japan's real estate prices peaked around 1991 (according to Richard Koo's presentation graphs) so perhaps the delay between peaking asset bubbles contributed to sustaining inflation. Nevertheless, there was disinflation (falling positive inflation), so in hindsight I should have aligned the different series for all these graphs via the peak year-over-year CPI rate of change rather than the peak price level itself.

The peak of Japan's CPI occurred in October 1998, almost eight years after the stock market peaked, and Japan's notorious mild deflation has been in effect since then:

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

Is the Great Depression experience or the Japanese experience more relevant for the US and the world today? I don't know the answer, but I would guess price inflation/deflation dynamics will be closer on the spectrum to the Japanese experience than to the Great Depression, though I think the dramatic price decline scenario could occur. With practically the entire world simultaneously awash in excess capacity and experiencing debt deflation (contraction of debt levels due to private sector deleveraging), the underlying conditions are certainly closer to those of the Great Depression, which was the last global debt crisis. But will developing countries such as China achieve sustainable growth in the next few years and provide a backdrop of growth much like that which surrounded Japan post-1990? On China at least I am skeptical.

The key question, to which most observers believe the answer is yes, is will structural differences in modern economies (versus the 1930s) or differences in government responses be able to avert severe broad price deflation this time? I sincerely hope the majority is correct for a change.

UPDATE 9/7/2009: A commenter pointed out that it may be too early to tell the effect of stimulus on core CPI, which is true. The data I've shown is not conclusive with respect to any future outcome. However, in the context of contracting private debt and the trends I've discussed before I'm predisposed to expect the types of outcomes described by the likes of Steve Keen (who builds on the work of Hyman Minsky among others) and Richard Koo (author of the phrase "Balance Sheet Recession").