Richard C. Koo, Chief Economist of the Nomura Research Institute and of Balance Sheet Recession fame recently wrote a paper entitled “Central Banks in Balance Sheet Recessions: A Search for Correct Response.”
This post is Part 1 of two posts in which I respond specifically to his remarks on Japan’s Lost Decade:
“These unusual phenomena are all caused by the fact that private sectors in all of these countries are massively increasing savings or paying down debt despite record low interest rates. According to the flow of funds data, the US private sector (household, corporate and financial sectors combined) today is saving whopping 6.9 percent of GDP (four-quarter moving average ending in Q4, 2012) at zero interest rates The comparable figure for the UK is 3.8 percent, for Ireland 8.6 percent, for Spain 8.1 percent and for Portugal 7.0 percent of GDP (Exhibit 2) all with record low interest rates. In Japan, where the bubble burst two decades earlier, the private sector is still saving 8.8 percent of GDP at zero interest rates (Exhibit 3).
Moreover, in all of the above countries, not only household sector but also the corporate sector is increasing savings or paying down debt at these record low interest rates. This behavior of the corporate sector runs totally counter to the conventional framework of neoclassical economics where profit maximizing firms are expected to be increasing borrowings at these record low interest rates.
Private sectors in all of these countries are increasing savings or paying down debt because their balance sheets were damaged badly when asset price bubbles burst in those countries. In the case of Japan, where the bubble burst in 1990, commercial real estate prices fell 87 percent nationwide (Exhibit 4), destroying balance sheets of businesses and financial institutions all over the country.”
The Bank of Japan has financial transactions data going back to 1964. Where more than one data series was available (1998Q1-1999Q1) I have used the more recent series. The following is a graph of calendar year averages of the financial surplus as a percent of GDP for the non-government sector (which includes public financial institutions and corporations), general government and the rest of the world:
As can be seen the non-government sector was in significant financial surplus from 1993 through 2005, averaging 8.9% of GDP. The largest swing towards surplus occurred between 1990 and 1998, going from 0.5% of GDP to 12.6% of GDP with both ratios being the extreme values for the calendar year averages.
Before moving on, note that the non- government sector was also in significant financial surplus from 1975 through 1986, averaging 7.9% of GDP. This was period of significant disinflation in Japan with consumer price inflation falling from 23.2% in 1974 to 0.6% in 1986. There were recessions in 1973-75, 1977, 1980-83 and 1985-86 and the IMF’s estimates of Japan’s output gap, which only start in 1980, show Japan below potential output through 1987, with the output gap in 1980 (6.5%) being nearly as great as in 2009 (6.8%). Aside from the absence of asset price bubbles the primary difference between this period and the more recent period is that nominal GDP (NGDP) growth rates, inflation and interest rates were higher, with the NGDP growth rate, GDP implicit price deflator inflation rate and the call rate averaging 8.1%, 3.7% and 6.9% respectively during 1975-86 as compared to 0.4%, (-0.8%) and 0.6% during 1993-2005. So interest rates were much higher, but nominal growth rates were also much higher, and thus the interest rate channel of the monetary transmission mechanism had plenty of space with respect to the zero lower bound.
The next graph shows the non-government sector financial surplus disaggregated between the household/nonprofit sectors (what the Japanese term the “personal” sector, and which itself is only available in disaggregated form for 1998Q1 on) and what Koo terms the “corporate” sector (the financial and non-financial sectors combined):
Here you can see very clearly what Koo thinks is the primary symptom of Japan’s Balance Sheet Recession, the shift by the corporate sector from a 9.2% deficit in 1990 to 9.8% surplus in 2003, or a total of 19.0% of GDP. Note that the corporate sector was always in financial deficit prior to 1995 and has almost always been in financial surplus since. Also note that the financial surplus of the household sector is smaller since 1996 than any time prior to that. Thus the corporate sector accounts for the entire non-government sector shift towards financial surplus from 1990-98.
Given the substantial structural differences between the balance sheets of the financial and non-financial sectors I find it curious that Koo lumps the two together under the “corporate” nametag. Here is a graph of the corporate sector financial surplus disaggregated between these two sectors:
Yes the financial sector ran both its largest financial surplus and deficit during the volatile years of 1998-2000, but mostly the financial balance of this sector is relatively small which explains why the non-financial corporate sector accounts for almost all of the changes in balance of the corporate sector. In fact the non-financial corporate sector shifted from a financial deficit of 8.9% of GDP in 1990 to a surplus of 2.8% of GDP in 1998 accounting for 95.7% of the non-government sector shift towards surplus between those years, and the shift to a financial surplus of 9.2% of GDP in 2003 means that the non-financial corporate sector accounts for all of the corporate sector shift towards surplus from 1990-2003.
Koo refers to the effect of the decline in land and share prices in his Exhibit 4 (“Cumulative Loss of Wealth on Shares and Real Estate ~ ¥1500 Trillion”). Stock data for Japan’s non-financial assets are available on a calendar and fiscal year basis from the Cabinet Office from 1969 and thus complete Japanese balance sheet data is only available at an annual frequency. Again, where multiple series are available (1980-1998 and 2001-2009) I have used the more recent series.
As mentioned there are substantial structural differences between the balance sheets of these sectors. As one might imagine the financial sector is much more leveraged with liabilities averaging 95.5% of assets over calendar years 1969-2011 compared to 72.5% for the non-financial corporate sector. Whereas land made up 32.2% of the non-financial sector’s assets at the peak of the Japanese land price boom in 1990, it was only 4.0% of the financial sector’s assets that year. Similarly shares made up only 5.7% and 7.7% of the financial sector’s assets and liabilities respectively at the end of calendar year 1989 (the Nikkei 225 peaked in December 1989). In fact since the financial sector held 138.0 trillion yen and 187.2 trillion yen as assets and liabilities respectively the decline in share prices likely helped to offset the financial sector’s losses due to the decline in land prices. In any case the decline in land and share prices apparently had little direct effect on the financial sector’s degree of leverage.
As long as we’re talking about the effect of the decline in land and share prices on balance sheets perhaps we should take a closer look at the “87%” figure for the decline in commercial land to which Koo refers. The index he is referring to is the Japan Real Estate Institute Urban Land Price Index for Commercial Land in Six Major Cities. “Six Major Cities refers to the ku-area of Tokyo, Yokohama, Nagoya, Kyoto, Osaka and Kobe. And while no doubt these cities are very important the Real Estate Institute also has All Urban and All Urban except Six Major Cities commercial land price indices that make it clear that the index for the 6 major cities is not very representative of the boom and bust in urban commercial land prices in Japan:
The All Urban commercial land price index, which is more representative of the price of the land that the non-financial corporate actually sector holds, has fallen from 195.5 in 1991 to 60.6 in 2005 or by 68.0% and evidently has been more or less stable since. This helps to explain the actual decline in the value of land on the asset side of the non-financial corporate sector balance sheet. The value of land held by the non-financial corporate sector fell from 634.2 trillion yen in 1990 to 281.8 trillion yen in 2011, or a difference of 352.4 trillion yen and a decline of 55.6%.
Shares made up 317.8 trillion yen or 16.4% of the non-financial corporate sector’s assets and 726.6 trillion yen or 37.5% of its liabilities in 1989. Thus shares were a net liability of 408.8 trillion yen and, as with the financial sector, a decline in share prices would probably have reduced the degree of leverage in the non-financial corporate sector. In fact shares were a 226.2 trillion yen net liability to the non-financial corporate sector in 2011 or 182.6 trillion yen less than in 1989. This offsets more than half of the decline in the value of land on the asset side of the non-financial corporate balance sheet. All told, the decline in the net value of nonfinancial sector land and shares assets and liabilities between 1989/1990 to 2011 is thus 169.8 trillion yen. Adding in an 11.7 trillion yen decline in the net value of financial sector land and shares assets and liabilities yields a loss of 181.5 trillion yen in corporate sector net worth, which although substantial, is substantially smaller than the 1500 trillion yen figure Koo mentions.
In any case, what I hope I’ve shown so far is that the shift in Japan’s non-government financial surplus by itself is not a new phenomenon, and that its more recent occurrence has been almost entirely due to the non-financial portion of what Koo terms the “corporate” sector. In addition I think it’s clear that Koo greatly exaggerates the damage to the Japanese corporate sector balance sheet from the decline in asset prices.
Another important question to answer is whether there is in fact any evidence that the degree of leverage has any relationship with the non-financial corporate sector financial surplus. Given we are talking about the supposed effect of an asset price decline induced increase in leverage on sector financial flows, there’s really only two simple coherent ways of measuring leverage in this context. The following is a graph of the ratio of liabilities to assets by sector:
Recall that the ratio of liabilities to assets averaged 72.5% in the nonfinancial corporate sector over 1969-2011. It’s true that this ratio was unusually high in 1989 and was above average during most of 1983-2007, but just eyeballing these graphs it’s hard to see much of a correlation between this measure of leverage and non-financial corporate sector financial surplus. Similarly, although non-financial corporate sector net worth as a percent of GDP declined from 124.9% in 1990 to 56.4% in 1999, and it was below its average value (96.9% over 1969-2011) during 1994-2000, and again from 2003-2007, there’s no obvious relationship between non-financial corporate net worth and non-financial corporate financial surplus.
Fortunately we don’t have to guess, we can always check for a statistical correlation. Here’s the graph of non-financial sector financial surplus regressed on liabilities as a percent of assets:
So since the state of non-financial corporate balance sheets doesn’t seem to explain the degree of non-financial corporate sector financial surplus very well, is there another explanation for this phenomenon that works somewhat better? Well, most of the non-government sector financial surplus is simply a residual of the government sector financial deficit. And since government revenue tends to fall and expenditures tend to rise in recessions it should come as no surprise that the government sector financial deficit is highly correlated to the output gap. Here’s the government sector financial surplus regressed on the output gap (as estimated by the IMF) from 1980 to 2012:
This is statistically significant at the 1% level. And it’s of greater statistical significance than the correlation between the non-government sector financial surplus and the output gap (which is only one of two residuals):
This is in turn is of far greater statistical significance than the correlation between the rest of the world financial surplus and the output gap (not shown). The regression of the government financial surplus on the output gap is also of far greater statistical significance than the regression of non-financial corporate sector surplus on the output gap (which has a p-value of 9.7%):
Bottom line: All of this is highly suggestive of the fact that Japan’s non-government sector financial surplus, which includes the non-financial corporate sector financial surplus, is far more a story about the government sector’s business cycle driven borrowing needs than of the private sector’s balance sheet repair driven savings needs.
In Part 2 I’ll discuss Koo’s opinions concerning Japan’s monetary and fiscal policy during the Lost Decades.