16 January 2016
Economists generally agree that the worst recessions of the past century are the Great Depression that began in 1929, the Japanese recession that began in 1991, and the Great Recession that began in December 2007. The three share several features in common. They were all preceded by massive asset bubble collapses, they were stubbornly resistant to the remedial efforts of governments and central banks, and they set economists debating about the fundamentals of macroeconomics. But in one important respect Japan's "lost decade" is completely different from the other two. Indeed, it does not even deserve to be called a recession.
That is a strong claim to make, and we hope the reader will be patient while we explain. A good place to start our analysis is the recession of 1945. In 1943 Paul Samuelson, then a young but rising economist, predicted that if the US war machine was suddenly wound up "there would be ushered in the greatest period of unemployment and industrial dislocation which any economy has ever faced." No one paid heed to Samuelson's warning and the US war effort was brought to an end very rapidly. The result was a 12.7% peak-to-trough drop in output. To put that in perspective, the Great Recession saw only a 4.3% drop in output. But none of Samuelson's dire predictions came to pass. The Dow Jones industrial index rose by 20%. There was no pileup in inventory. Companies managed to sell all they could produce. In fact demand exceeded supply and inflation ran at 2.3%. Unemployment was around 5%.
The explanation is involved. During the war, through a combination of rationing and the sale of war bonds, forced personal saving rose to around 25% from less than 5% before the war, so pent-up savings were high at the end of the war. But it was not that people ran down their accumulated saving during the 1945 recession. If that were the case then the saving rate would have turned negative, which it never did. It was rather that people cut their saving rate sharply though it still remained higher than the level before the war. So although output and thus personal income fell by more than 12% this loss in aggregate demand was compensated by the fall in the saving rate. Thus aggregate income fell but not consumption demand. The 1945 recession was a GDP recession, and a very severe one at that, but it was not a consumption recession. There was no pain. That is why the history books refer to it as "technically a recession". My book Macroeconomics Redefined has a detailed model of what happened during the war and recession that followed.
The 1945 recession is a warning that in order to gauge the severity of a recession economists should be looking at consumption instead of GDP. The graph below shows year-on-year changes in real personal consumption expenditure in the US from 1960 to 2015. To calculate real personal consumption expenditure I have used Personal consumption expenditure from the St Louis Federal Bank site and adjusted for inflation using the consumer price index for all urban consumers. What is readily seen is that the recessions of 1960-61 and 2001 were very mild and hardly deserved to be termed recessions from the viewpoint of consumption. The 1973-75 recession, the Volcker recessions, and the Great Recession were very severe. The 1991 recession was moderately severe.
When Japan's equity and real estate bubbles burst around 1991, the principal blow fell on companies and banks. Consumers were comparatively unaffected. Companies that took hefty hits to their balance sheets responded by cutting output and investment. This meant a loss in aggregate income. But consumers responded by cutting their saving rate, which fell from 13.3% in 1991 to 6.3% in 2003 [The Causes of Japan's 'Lost Decade': The Role of Household Consumption by Charles Yuji Horioka, March 2006, National Bureau of Economic Research], and continued to fall thereafter. Unlike in the US during the Great Recession, where consumers raised their saving rate and thus caused a fall in aggregate demand, Japanese consumers mitigated the fall in aggregate demand (caused by a fall in investment and output) by cutting their saving rate. As a result, household consumption expenditure as a percentage of GDP rose from 52.82% in 1991 to 61.14% in 2013. (data from Index Mundi: Household final consumption expenditure in constant yen)
The graph below shows that the YoY growth rate of real consumption became negative only briefly in 1998, seven years after the crash, and then again during the 2008-09 global recession.
Upto this point the Japanese experience followed the pattern of the 1945 recession in the US. But there the parallel ends. US company balance sheets improved during the course of the war. So they could respond to increased consumption demand by stepping up investment. In Japan companies found their balance sheets ravaged by the combined collapse of the equity and real estate bubbles. So even if they wanted to increase investment they could not because of the sheer absence of money. In Japan the fall in aggregate demand was caused by a fall in investment demand, not principally a fall in consumption demand. In the Great Depression and the Great Recession the fall in aggregate demand began with a fall in consumption demand to which companies responded by cutting investment sharply. This is reflected in the fact that while the household saving rate in Japan fell sharply, the corporate saving rate rose substantially as companies tried to repair the large hole in their balance sheets.
When one understands the mechanism of the Japanese recession it is also easy to understand why neither the huge fiscal spending nor the cut in interest rates to near zero levels had much impact. Fiscal spending works by reversing the fall in personal incomes (and thus consumption demand) caused by a fall in investment demand. In Japan there was never a sharp fall in consumption demand because the high saving rate enabled Japanese consumers to counter the fall in personal income by cutting their saving rate. So although pouring millions of tons of concrete put money into consumers' hands it did not help companies to raise their investment.
Monetary policy was ineffective for the same reason. A cut in interest rates works by cutting the cost of capital to below the expected return on capital and thus making more projects viable. But in Japan there was never a substantial fall in the return on capital. Consumers had all along been willing to buy whatever was available to be bought. The problem for firms was not the cost of money so much as the sheer non-availability of money. Companies had no money because they had lost gargantuan amounts in the crash. And they could not borrow because banks had been fellow victims of the crash.
The one piece of advice the Japanese central bank should have followed was that of Walter Bagehot, tendered in the 19th century: "Lend freely, at a high rate, on good collateral." But because it was not backed by the elegant mathematical models of the kind now favoured by economists no one paid it any heed.
11 January 2016
The St Louis Federal Bank web site has a paper entitled Why Did Loan Growth Stay Negative So Long after the Recession? by Maximiliano Dvorkin and Hannah Shell.
The first graph on it shows that QoQ loan growth turned negative soon after the start of the 2007 recession and stayed negative for a long time thereafter. This was in sharp contrast to the recessions of 1990-91 and 2001. The paper concludes that both the demand for loans and the supply of loans played a role in this behaviour.
One way of estimating the demand for loans is to graph the movement of Real Personal Consumption Expenditure. Below I have drawn a graph of Real PCE for the three recessions and it is noteworthy how closely the line for the 2007-09 recession follows the graph of loan growth in the above-mentioned paper compared with the previous two. It seems that the 2007-09 recession was much more a demand-driven recession compared with the others.
Why should Real PCE growth be related to the demand for loans? The reason is that capitalists make investments depending on the change in consumption compared with the previous time period. Remember that in Paul Samuelson's famous accelerator paper of 1939, I(t) is a function of C(t)-C(t-1).