Balance sheet crises: causes, consequences, and responses.

• Author: Gjerstad, Steven

Being the managers rather of other people's money than of their own, it cannot well be expected, that they should watch over it with the same anxious vigilance with which [they] frequently watch over their own.

--Adam Smith

Balance sheet crises, in which the prices of widely held and highly leveraged assets collapse, pose distinctive economic challenges. An understanding of their causes and consequences is only recently developing, and there is no agreement at all on effective policy responses. A preliminary purpose of this article is to examine in detail the events that led to and resulted from the recent U.S. housing bubble and collapse, as a case study in the formation and propagation of balance sheet crises. The primary objective of the article is to evaluate similar events around the world with a view toward assessing the economic performance of countries that have pursued varied alternative policies.

We propose that the Great Depression beginning in 1929 and the Great Recession starting in 2007 were both household-bank balance sheet crises--events that were quite distinguishable from the recessions appearing between them. Each episode, we hypothesize, was preceded by unsustainable rises in expenditures on construction of new housing units and in mortgage credit for purchases of new and existing homes. In both cases housing values rapidly collapsed by, more than 30 percent but mortgage debt obligations fell only very slowly, so that housing equity fell sharply. (1)

Between these two economic calamities were 12 smaller recessions. Nine of the ten recessions between World War II and the Great Recession were led by declines in new housing expenditures and in all of those the interaction between Federal Reserve monetary policy and the housing-mortgage market was a clearly discernible feature. Federal Reserve monetary policy between the fall of 1979 and the summer of 1982 is a prominent example of this interaction effect, and an excellent "natural experiment" on the impact of monetary policy on the mortgage and housing markets. Examination of the normal impact of monetary policy and the contrast with economic conditions in the aftermath of the housing bubble suggests why monetary policy has had so little effect on the money supply and the economy over the past five years. When households are awash in debt, banks face a continuing legacy of impaired assets mad damaged balance sheets, and there is a large inventory of unsold and foreclosed homes hanging over the housing market, low short-term interest rates don't stimulate lending to nearly the extent that they do in normal times. Consequently, monetary policy doesn't have its normal effect during a balance sheet crisis. We also provide direct evidence from other countries that fiscal stimulus has not been a part of the recovery process in many countries that have had robust growth soon after a balance sheet crisis. In fact, most countries that have recovered rapidly have first contended with a rapid increase in government deficits but have soon reduced both government expenditures and government deficits.

Widely differing approaches have been taken regarding the recognition of losses on the impaired assets of financial institutions. One approach is to shore up financial institutions and allow them to slowly recognize past losses. The opposite approach is to force lenders to recognize losses, even to the point of wiping out equity holders and forcing "haircuts" on bondholders. We evaluate the effects of confronting balance sheet problems, especially in financial institutions, by contrasting Sweden, which aggressively addressed the impaired conditions of its banks, with Japan, which allowed its banks to stretch out recognition of losses on bad assets for over a decade. Sweden recovered quickly while Japan languished for over a decade.

Market currency depreciation is a prominent feature of recovery in many countries. We discuss and chart three disparate examples--Finland, Thailand, and Iceland--that illustrate and explicate their significant, and common recurring, features.

Fiscal responses have also differed widely. Most countries that have experienced an asset market bubble and collapse have incurred deficits of about 10 percent of GDP per year afterward--due to declining revenues and increased expenditures--to address the financial crisis and loss of income. Japan is an extreme example of Keynesian deficit spending that has continued for well over a decade 'after the crisis began. At the other extreme, some countries have brought deficits below 2 percent of GDP within two or three years. The results have almost uniformly favored the countries that have controlled their deficits.

Housing and Mortgage Credit in the Depression

Aside from the large inflows of foreign capital that supported the housing bubble during the Great Recession, the development of the Great Depression and the course of the Great Recession included many common elements. In the next two sections, we demonstrate the parallels between the housing expansion and early collapse phases of the two episodes.

Proposition 1: The Great Depression and Great Recession were disequilibrating housing market and mortgage credit booms.

Figure 1 plots annual observations from 1922 through 1937 for the following expenditures: gross national product (GNP), nondurable consumption spending (C), consumer durable goods spending (D), nonresidential fixed capital investment (I), and new housing construction expenditures (H). As a vehicle for conveying the relative changes in each of the various measures, each point is plotted as a percentage of its level in 1929, the year the Great Depression started. (2)

[FIGURE 1 OMITTED]

Most notably, a major housing investment collapse preceded the Great Depression by over two years yet all other major expenditure categories continued to rise. Housing expanded rapidly by nearly 60 percent from 1922 to 1925, leveling out in 1926 and then began its long descent, not bottoming out until 1933. In 1929 new housing expenditure had returned to its 1922 level before any of the remaining expenditure categories had declined more than small temporary amounts. GNP and each of its major components declined in 1930. Uncharacteristically, as recessions play out, even nondurables declined, although less steeply than every other category of expenditure.

Figure 2 shows the net flow of mortgage funds from 1900 to 1940. The solid curve is the exponential trend growth of mortgage lending from 1897 through 1922. The dashed curve is the extension of the trend forward into the boom years of the 1920s and into the collapse during the Great Depression. In the residential mortgage lending data that we have from 1897 through early 2013, the only collapse like that of 1929-33 came in the second quarter of 2006 and persisted through the first quarter of 2013 at low negative net flow rates.

[FIGURE 2 OMITTED]

The initial decline in housing construction from 1927 to 1929 (Figure 1) preceded the sharp decline in the net flow of mortgage credit beginning in 1929, indicating that the contraction in new housing expenditures came well in advance of the reduced flow of credit into the housing market. The suspended two-year lag of mortgage credit behind the decline in the rate of investment in new house construction implies increasing leverage for home purchases in 1927 to 1929. Gjerstad and Smith (forthcoming) provide other disaggregated evidence of increasing leverage in the later years of the 1920s boom.

Comparing Figures 1 and 2, the key feature that requires emphasis is that the 60 percent increase in the rate of new home construction expenditures from 1922-25 was matched by a 200 percent increase (from $1 billion to $3 billion) in the net flow rate of mortgage credit. Moreover, this large net credit flow continued through 1928 before it collapsed. Consistent with a relatively elastic supply of new home construction in the 1920s, the housing price data available from the years leading up to 1925-26 (before expenditures start to fall) do not show all increase comparable to the recent price bubble. But we suggest that large house price increases are not a necessary condition for severe subsequent household-bank balance sheet stress. A more elastic supply means that more units are being added to the stock to be impacted by a price turn, even if the impact on each unit is smaller--that is, more balance sheets will be in distress although each will be less severely stressed than if the price run-up had been larger and housing output smaller. The large home price decreases came after 1930.

Housing, Mortgage Credit, Foreign Capital Inflows, and the Great Recession

Much of the above Great Depression narrative was, and is in process of being, repeated in the Great Recession. The year 2012 marks the fifth year since the recent downturn, corresponding to 1934 in Great Depression clock time. Figure 3 charts the same measures as Figure 1, except that we report GDP for the Great Recession rather than the GNP measure of aggregate output that we reported for the Great Depression. (3) We also drop nondurable consumption (C) from the chart since its behavior so characteristically follows a damped version of GDP in a slump. (4) And we have added a plot of unit sales (S) of new homes. Again, housing provided substantial lead time for the impending recession that began in the fourth quarter of 2007: expenditures on construction of new homes peaked seven quarters before the recession began, and unit sales peaked nine quarters before the recession. The devastation that followed is apparent in that it was not until the third quarter of 2011 that GDP recovered to its recession peak--the longest GDP downturn in the United States during the post WW II period.

[FIGURE 3 OMITTED]

Unit sales of homes declined while expenditures continued to rise because prices and the flow of credit continued to rise unabated. Builders tend to...