II Recession and Recovery in the 1990s

Ramana Ramaswamy; J. H. Green

Jump to Content Jump to Main Navigation

Author: Mr. Ramana Ramaswamy and Mr. J. H. Green

Language:: English

Keywords:: OP; recession; labor market program; pension; investment ratio; housing investment; consumption boom; recession to the stance; nondurable consumption; Private consumption; Consumption; Real interest rates; Vector autoregression

Download PDF (2.1 MB)

Abstract

After almost a decade of expansion in the 1980s, the Swedish economy entered into a deep recession beginning in the first quarter of 1990. The cumulative decline in output during this recession, which lasted a full three years, was around 7 ½ percent. This makes the recent recession the longest and deepest since the 1930s. Moreover, it contrasts sharply with the previous recession of 1980-81, which lasted barely one year and during which output declined by only about 2 ¾ percent (Chart 2-1). Since the second quarter of 1993, a significant, albeit unbalanced, recovery has been under way, driven in the initial stages primarily by a major strengthening of the external sector.

This section examines the proximate causes of the recession of the 1990s, the reasons for its long duration, and the impact that it had on the economy. The section then compares the recovery that is currently under way with the recovery in the early 1980s. Finally, a vector autoregression analysis provides a more rigorous underpinning to the analysis of the recession and recovery in the 1990s. The analysis indicates that the proximate cause of the recession in the 1990s can be ascribed in large measure to demand shocks—particularly shocks to private consumption and residential investment.

Proximate Causes of the Recession

The proximate (or immediate) cause of the recession in 1990 could have come from any of a variety of sources. Among the more likely candidates are (1) changes in the stance of fiscal policy; (2) shocks from the external sector; (3) the stance of monetary policy preceding the recession; and (4) shocks to the individual components of demand.

As can be seen from Chart 2-2, which presents a measure of the fiscal impulse over the past decade, fiscal policy was expansionary when the economy went into recession.¹ Indeed, the fiscal impulse measure was positive from 1988 onward, and, in fact, became strongly so during 1990-91. This expansionary stance until 1993 is a reflection of several factors, including the impact of the 1990-91 tax re-form and the large increase in real household transfers during this time.² Thus, if anything, fiscal policy can be considered to have had an expansionary effect on activity both prior to and during the recession, and hence should be excluded as a proximate cause of the downturn.

Chart 2-3 shows that the external sector’s contribution to growth had been consistently negative in the latter half of the 1980s, because of the adverse consequences of the cumulative decline in competitiveness during this period.³ However, there was no worsening in the external sector’s negative contribution immediately preceding the recession, while with the onset of the recession the external sector’s contribution became distinctly positive. Furthermore, since 1993 the external sector’s contribution to growth has been strongly positive, and it is mainly responsible for the turnaround in overall output growth since the second quarter of 1993. These considerations would appear to exclude the external sector as a proximate cause of the 1990-93 recession.

Chart 2-2 also indicates the relative stance of monetary policy over the past decade. While the Riksbank increased its marginal lending rate from around 9 percent in the first quarter of 1988 to a high of over 14 percent in the last quarter of 1990, real interest rates did not increase significantly during this period as inflation accelerated. In fact, the real monetary conditions index (a weighted average of the real interest rate and the real exchange rate) remained broadly unchanged between 1987 and 1990. However, real interest rates rose sharply toward the end of 1991 and particularly in the third quarter of 1992, as nominal interest rates were raised dramatically in an attempt to defend the currency. The substantial tightening of monetary policy that preceded the floating of the exchange rate led to a record level for the monetary conditions index (Chart 2-2). Given the lags involved in the monetary transmission mechanism, although it is difficult to attribute the proximate cause of the recession to the stance of monetary policy, it would appear plausible that the abrupt monetary policy tightening in 1992 and the related uncertainties in financial markets may very well have substantially deepened and prolonged the recession.⁴

This leaves domestic demand shocks as the most likely proximate cause of the most recent recession. Chart 2-3 shows that the trends in the personal savings and investment ratios are indeed consistent with this hypothesis. Thus, starting in the first quarter of 1990 the household savings ratio rose sharply and the investment ratio declined steeply. The question that arises is why the demand shocks occurred at this particular juncture. This question is addressed at an intuitive level in the next section and more rigorously below using the technique of vector autoregression.

Causes of Demand Shocks

Consumption

The sharp contraction in domestic demand from early 1990 is very much related to the nature of the boom in private consumption that followed the financial liberalization in 1985. Following the liberalization, households could borrow more freely and financial institutions seeking to increase their market share lent more aggressively. With the easier availability of credit, households increased their borrowing substantially to meet their demand for consumer durables and housing, which was pent up by the earlier episode of credit rationing.⁵ Consequently, household debt as a percentage of disposable income increased from about 100 percent in 1985 to 130 percent in 1988 (Chart 2-4). However, with asset prices booming during this period (property prices increased by 50 percent between 1985 and 1990), households could finance their increased borrowing on the basis of rising net wealth (Chart 2-4). As a result of this process, households began to dissave, and there was a rapid growth of private consumption. This growth in consumption was fueled largely by durables, with purchases of automobiles alone increasing at almost 30 percent a year between 1985 and 1987 (Chart 2-5).

The consumption boom started to dissipate from 1989 onward, and private consumption declined in 1990, precipitating the recession. The shock to private consumption in 1990 came as the culmination of a number of different developments. First, posttax interest payments rose as the tax reform of 1990-91 reduced the tax deductibility on interest payments (Chart 2-4). As this was combined simultaneously with higher levels of debt and falling asset prices, households embarked on a process of major balance sheet adjustment. In addition to the need for financial consolidation, the marked increase in the private savings ratio was also conditioned by precautionary factors, as the future direction of the welfare state became increasingly uncertain during this period. In consequence, household savings in relation to disposable income swung dramatically from a negative 4 percent in 1989 to a record 8 percent in 1992 (Chart 2-4). This had the effect of precluding any early revival of private consumption demand and of prolonging the recession.

Investment

Gross fixed investment grew at a rapid rate following the financial liberalization, increasing in real terms by about 12 percent in 1989. The subsequent decline in fixed investment was especially sharp—the cumulative decline during 1991-93 was about 35 percent (Chart 2-5). The rapid growth of investment in the late 1980s was influenced in part by the liberalization of the financial markets and in part by the generous subsidies to the housing sector. While the financial liberalization in 1985 relieved insurance companies from the obligation to invest in government and housing bonds, they were not given the freedom to invest abroad until 1989. In the interim, given that the financial liberalization was not accompanied by monetary or fiscal tightening, insurance companies invested heavily in commercial real estate, contributing to the boom in property prices.

The construction boom ended in late 1990, as inflationary expectations came down substantially and there was a supply-side squeeze on credit for real estate investment. Business investment was also cut back substantially because of falling demand, rising debt, declining profitability, and increases in real interest rates (Chart 2-5).⁶ The absence of any significant revival in fixed investment until 1994 is yet another factor that contributed to the length of the recession.

Selected Features of the Recession

The output and employment losses during the past recession were substantial. Between 1991 and 1993, the cumulative loss in aggregate output was approximately 5 percent, while aggregate employment fell by 12 ½ percent during the same period. The impact of the recession on the labor market was particularly calamitous, with total unemployment (which includes participants in government-sponsored labor market programs) increasing from just over 4 percent of the labor force in 1990 to almost 13’A percent in 1993.⁷ The increased gaps in the product and labor market had, however, the favorable effect of sharply reducing Sweden’s high wage and price inflation. In particular, consumer price inflation fell sharply from an average of over 10 percent in 1990 to about 2 ½ percent in 1992 (Chart 2-6).

The recession resulted in an especially sharp decline in manufacturing output, owing to reduced levels of domestic demand, declining external competitiveness, and the overall weakness in trading partner economies (Chart 2-7). The decline in manufacturing employment was particularly sharp, falling by almost 25 percent between 1990 and 1993, as firms radically reorganized their production structures to adapt to the new and more difficult environment. As a result, there was a rapid increase in the growth of manufacturing productivity, which reached over 9 percent in 1993 (Chart 2-7).

The major decline in construction output occurred only after the recession was already well under way. The expansion of the construction sector in 1990 was mainly on account of the accelerated completion of existing projects following the expectation that value-added taxes would soon be increased on construction activity to a new unified level. However, the subsequent decline in construction output was particularly steep, falling by almost 9 percent in 1993 (Chart 2-7). The decline in construction employment, while initially lagging manufacturing employment, was steeper as the recession got under way, and the cumulative decline in construction employment was almost 20 percent in 1992-93 (Chart 2-7).

The services sector was not particularly adversely affected during the 1980-81 recession. In contrast, both output and employment in this sector fell substantially during the most recent recession, reflecting both the widespread impact of the recession, as well as the effects of attempting to control public consumption (Chart 2-7).

Comparing Economic Recoveries

As already mentioned, the recent recession lasted far longer and was much deeper than the recession in 1980-81. Furthermore, unlike the previous recession, which primarily affected the manufacturing sector, the impact of the 1990-93 recession was wide ranging, with employment in the services sector declining for the first time in many years. This section contrasts the main features of the current recovery with the one following the 1980-81 recession (see Charts 2-8 and 2-9).

A comparison of the behavior of private savings in the two cycles is revealing. The recessions in both 1980 and 1990 were precipitated by consumption shocks—reflected by the initial sharp increase in the household savings rate (Chart 2-8). However, although the upturn in the cycle of the early 1980s was mainly supported by the rapid growth of exports following the devaluations of 1981-82, it was also driven by the relatively quick recovery in private consumption. In fact, the household savings ratio started declining from the last quarter of 1981, contributing to a balanced recovery in the initial stages.

The beginnings of the recovery in the latter half of 1993 has been propelled mainly by the external sector. Export volume growth picked up to 15 percent in the year ended mid-1994, following the almost 30 percent increase in competitiveness due to the combined effects of the depreciation, rising labor productivity, and muted wage costs. However, unlike the previous cycle, following the onset of the recession in the beginning of 1990, the household saving ratio rose for 14 consecutive quarters (Chart 2-8). Furthermore, even the weak recovery in private consumption toward the end of 1993 was dampened by rising interest rates and growing uncertainties about prospective tax increases.

During the previous cycle, the gross fixed investment ratio continued to fall even after output started recovering (Chart 2-8). Investment began to grow only in the later stages of the cycle. In contrast, fixed investment—primarily in the business sector—has picked up in the early stages of the current recovery and is expected to grow by 18 percent in 1994. Housing investment, however, continues to fall steeply. The strong performance of business investment in the early stages of the recovery is mainly on account of the boom in exports. Unlike private consumption, business investment has been less constrained by rising interest rates in 1994, as the depreciation saddled exporting firms with large profit margins.

A Vector Autoregression Analysis

A vector autoregression (VAR) approach lends itself to a more technical means for capturing and quantifying the various demand shocks to the macroeconomy. By estimating a system of equations, in which each variable in the system is determined by its own lagged values and the lags on all other variables in the system, one can identify shocks that are not easily explained by either the past history of that variable or the history of other variables which impact on it. For instance, a consumption shock can be identified as those changes in private consumption that are not strictly related to changes in income, wealth, or the actual stance of policy. Rather, such a consumption shock might arise as a consequence of uncertainty or pessimism that people have about the future course of policy. The identification of such shocks is particularly pertinent in the case of Sweden, as there was increased uncertainty during this period about the basic direction of policies that might affect future societal arrangements.

The VAR model estimated for Sweden includes the following as endogenous variables: private consumption (both durables and nondurables); investment (both residential and nonresidential); changes in inventories; and imports. The remaining items of the GDP identity—government consumption and exports—are considered exogenous to the system. The model also includes monetary policy variables, which are assumed to be exogenous to the system. These include the real interest rate and the real exchange rate. In view of the critical role of household savings during the recession, and the potential link to balance sheet adjustment, net household wealth is also included as an exogenous explanatory variable in the VAR system.

The estimation of the VAR system for Sweden uses the standard approach in which each endogenous variable is regressed on past values of all endogenous variables and the exogenous variable in the system.⁸ For the current exercise, a shock can be identified as a persistent pattern of negative or positive forecast errors for a variable. For example, a persistent pattern of negative residuals in the consumption equation would represent a negative shock to private consumption. That is, a series of negative errors in the consumption equation can be taken as evidence of the fact that changes in private consumption are not being explained by the history of past changes in income or other factors that affect consumption, but instead are predicated on uncertainty or pessimism.

Table 2-1 reports the results of the VAR estimated over the period 1970:Q1 to 1994:Q2. Each observation in the table represents the sum of shocks cumulated from the starting point of the recession in the first quarter of 1990. As mentioned above, each shock is defined as the residual of the equation, normalized by its standard deviation, so that individual residuals are distributed with mean 0 and unit standard deviation.

Table 2-1.

VAR Shocks (Cumulated from 1990:Q2)

Table 2-1.

VAR Shocks (Cumulated from 1990:Q2)

	GDP	Durable Consumption	Nondurable Consumption	Residential Investment	Nonresidential Fixed Investment
1990:Q1	0	0	0	0	0
1990:Q2	-0.2	-0.9	-0.2	1.6	-0.1
1990:Q4	-0.6	-3.1	-1.6	1.3	0.0
1991:Q2	-1.6	-2.2	-0.8	1.4	-0.1
1991:Q4	-3.6	-2.0	-1.8	1.6	-0.1
1992:Q2	-5.1	-3.8	-2.7	1.7	-0.4
1992:Q4	-5.4	-5.7	-3.0	0.4	-0.5
1993:Q2	-5.8	-5.1	-3.0	-1.8	-0.5
1993:Q4	-6.5	-5.1	-2.7	-5.4	-0.7

Table 2-1.

VAR Shocks (Cumulated from 1990:Q2)

	GDP	Durable Consumption	Nondurable Consumption	Residential Investment	Nonresidential Fixed Investment
1990:Q1	0	0	0	0	0
1990:Q2	-0.2	-0.9	-0.2	1.6	-0.1
1990:Q4	-0.6	-3.1	-1.6	1.3	0.0
1991:Q2	-1.6	-2.2	-0.8	1.4	-0.1
1991:Q4	-3.6	-2.0	-1.8	1.6	-0.1
1992:Q2	-5.1	-3.8	-2.7	1.7	-0.4
1992:Q4	-5.4	-5.7	-3.0	0.4	-0.5
1993:Q2	-5.8	-5.1	-3.0	-1.8	-0.5
1993:Q4	-6.5	-5.1	-2.7	-5.4	-0.7

The results show the buildup of negative shocks to both durable and nondurable consumption in the initial stages of the recession. This provides support for the hypothesis that consumption shocks were primarily responsible for precipitating the recession. The shocks to durable consumption dissipated to some extent in 1991, but then became very large in 1992, reflecting the high degree of uncertainty related to the financial and banking crisis. This had the effect of prolonging the recession.

The relatively smaller shocks to nondurable consumption compared with shocks to durables may be related to the large increases in real public transfers to the household sector during this period. These transfers, which resulted in part from the lagged indexation of most benefits in an environment of falling inflation, may have helped households to smooth their day-to-day consumption while raising their savings rate at the same time.

An interesting result that emerges from the VAR exercise is the absence of shocks to fixed investment. In fact, there were small positive shocks to residential investment in the early stages of the recession. This indicates that the decline in nonresidential investment is explained relatively well by the VAR system. That is, the changes in fixed investment during the recession were largely a response to accelerator-type effects; the initial consumption shock led to a decline in GDP, which, in turn, had a negative impact on fixed investment.

The positive shocks to residential investment during the recession are explained by specific housing policies, which are not captured by the model. In 1990, the housing market was supported partly by the anticipation of reductions in tax preferences in the 1991 tax reform. Residential construction projects were further supported in 1991 and the early part of 1992 by the anticipation of cuts in housing subsidies that were made necessary by the fiscal crisis.

The VAR model does not provide clear-cut evidence for the role of monetary policy in precipitating the recession—the shocks do not become bigger when the system is estimated without the monetary policy variables. This is consistent with the behavior of the monetary conditions index prior to the recession as noted above. However, the uncertainty surrounding the financial crisis in 1992 is most likely to have contributed to prolonging the recession.

The fiscal impulse measure, defined as the change in the structural balance, provides one particular indicator of the impact of fiscal policy on activity. A positive value for the fiscal impulse indicates an expansionary fiscal stance, and a negative value a contractionary one. See Heller, Haas, and Mansur (1986) for a discussion of the fiscal impulse measure.

See Section IV for a fuller discussion of the reasons underlying the expansionary stance of fiscal policy during the early 1990s.

Recent external developments are reviewed in Section VII.

A more rigorous analysis of the impact of monetary policy on the recession is undertaken below using a vector autoregression analysis.

The Swedish experience in this regard is similar to that of other countries that went through a process of financial liberalization. For a fuller discussion, see Muellbauer (1994) and Berg (1994).

Corporate debt increased from about 115 percent of GDP in 1985 to over 190 percent in 1990.

The consequences of the crisis on labor market developments are discussed more fully in Section III.

All variables, except as noted below, are in constant prices and in log difference terms to account for stationarity considerations. The change in stocks and the fiscal balance are entered as a share of GDP. The model uses three lags. In the VAR framework, variables are included in the model if they pass statistical tests for predicting other model variables. A variable is considered to be endogenous to the system if it is predicted by a variable other than itself; otherwise, it is assumed to be exogenous. For a more detailed discussion of the VAR methodology, and the identifying assumptions used for the estimation, see Catao and Ramaswamy (1995).

Abstract

Proximate Causes of the Recession

Causes of Demand Shocks

Consumption

Investment

Selected Features of the Recession

Comparing Economic Recoveries

A Vector Autoregression Analysis

Cited By

Share Link

Topics

Countries

Series

About

Resources

Abstract

Proximate Causes of the Recession

Causes of Demand Shocks

Consumption

Investment

Selected Features of the Recession

Comparing Economic Recoveries

A Vector Autoregression Analysis

Cited By

Within Same Series

Other IMF Content

Other Publishers Content

Asian Development Bank

Inter-American Development Bank

International Labour Organization

The World Bank

Share Link

Table of Contents

Sections

Figures

References

Metrics