Abstract

Hong Kong SAR has experienced continued deflation since the last quarter of 1998. The composite consumer price index (CPI) fell by almost 14 percent from the third quarter of 1998 until December 2002; about half of this deflation is accounted for by the decline in housing costs, following the bursting of the bubble in property prices that had built up in the mid-1990s.1 Other items that have also contributed significantly to the decline in prices include food, clothing and footwear, and durable goods (Table 5.1). Falling prices have contributed to increased real debt burdens, depressed consumer confidence, and tightened monetary conditions, and could have helped feed the contraction in aggregate demand (Table 5.2).

Hong Kong SAR has experienced continued deflation since the last quarter of 1998. The composite consumer price index (CPI) fell by almost 14 percent from the third quarter of 1998 until December 2002; about half of this deflation is accounted for by the decline in housing costs, following the bursting of the bubble in property prices that had built up in the mid-1990s.1 Other items that have also contributed significantly to the decline in prices include food, clothing and footwear, and durable goods (Table 5.1). Falling prices have contributed to increased real debt burdens, depressed consumer confidence, and tightened monetary conditions, and could have helped feed the contraction in aggregate demand (Table 5.2).

Table 5.1.

Contributors to Deflation

article image
Sources: Data from CEIC Data Co., Ltd. database; and IMF staff calculations.

Based on log-difference approximation.

CPI denotes the consumer price index. Sum of the components using fixed weights over the period September 1998 through December 2002.

Table 5.2.

Some Key Economic Indicators:

(Year-on-year percentage change, unless otherwise indicated

article image
Sources: CEIC Data Co., Ltd. database; Hong Kong SAR authorities; and IMF staff calculations.

Best lending rate minus actual year-on-year consumer price index (CPI) inflation.

Interbank offered rate minus actual year-on-year CPI inflation.

Levels of unemployment rate prevailing in 1997: Q3 and 2002: Q4.

Previous research has suggested that although structural factors have played a role, deflation was mainly attributable to cyclical factors. Analysis carried out by the IMF staff during the 2002 Article IV consultation with Hong Kong SAR showed that fluctuations in unemployment, nominal credit, and the nominal effective exchange rate—which were considered cyclical variables—were the main determinants of the decline in prices (IMF (2002), Chapter III). This result implies that as these cyclical factors turn around, deflation will end. Research by the Hong Kong Monetary Authority (HKMA, 2001) arrived at similar conclusions.

The persistence of deflation for such an extended period of time suggests, however, that other factors may be in play. In general, an economy’s cyclical condition, as measured for instance by the output gap, can be affected by both transitory and permanent shocks, the relative importance of which determines the persistence of deflation. For example, as deflation increases the real debt burden and becomes entrenched in expectations, private investment and consumption growth could decline, causing an even greater decline in economic activity.

This section presents a more comprehensive analysis, which provides a decomposition of the aggregate price level into transitory and permanent components, and identifies the nature and origin of the shocks that drive these two components. The analysis is based on a methodology with several features that are useful for analyzing deflation and its persistence. First, it provides a clear distinction between those driving forces behind deflation that create trend movements in the variables (summarized in the permanent component) and those that generate temporary deviations from long-run equilibrium conditions (summarized in the transitory component). Second, it provides a framework for identifying the nature of those forces. The approach undertaken here is a “structural” one, as opposed to the commonly used reduced-form approach. It helps, for example, to determine whether falling prices result from one or more of the following: increased productivity, scarce money supply, and temporary excess capacity. This is particularly relevant, because the likely duration of deflation, its costs, and the policy actions that may be needed to combat it all depend upon the nature of the deflation’s underlying causes.

Empirical evidence suggests that the contribution of the permanent component has become relatively more important over time in explaining deflation. The sustained fall in the aggregate price level is mostly accounted for by continuous declines in its permanent component, which summarizes the cumulative effects of productivity shocks, scarce money supply, and price convergence with trading partners. These findings indicate that although the transitory component did contribute significantly to the initial phase of deflation, its effects are becoming progressively weaker.

Framework

A structural vector error-correction-modeling approach à la King and others (1991) is used to assess the nature and impact of shocks on prices to shed light on the main factors that are behind their sustained decline. The structural vector error-correction model, also known as the common-trends model (CTM), is well suited for analyzing the interaction between variables that display trends and are determined simultaneously, uses general restrictions derived from economic theory to identify the main driving forces behind the trends observed in aggregate macroeconomic variables, and ensures consistency between the short-run and long-run dynamics of those variables.

The CTM includes the following variables: real output, measured as nominal GDP deflated by the (composite) consumer price index; broad money; the consumer price index; real asset prices, measured by the Hang Seng stock index deflated by the consumer price index; and foreign prices in Hong Kong dollars (HK$), measured as the trading partners’ consumer price indices expressed in HK$.2

Figure 5.1 provides a diagrammatic representation of this framework. It shows how the outturn of the price level can be decomposed into two components that reflect its short-run and trend movements, and that can be used to identify their main driving forces.

Figure 5.1.
Figure 5.1.

Effects of Shocks on Prices

Note: PPP denotes purchasing power parity.
  • Panel I displays several possible factors that could, at each point in time, influence different sides of the economy (for example, supply versus demand and nominal versus real).

  • The occurrence of an event related to these factors constitutes a “shock.” The nature of the shocks is determined according to which side of the economy they first have an impact on, as well as the temporal nature of the shock (see Panel II). There are two types of shock: transitory shocks, which generate only short-run movements in the variables, and permanent shocks, which generate both short-run and trend movements in the variables.3 The transitory shocks are cost-push shocks (for example, changes in markup margins), aggregate-demand shocks (for example, temporary shifts in consumers’ preferences), and liquidity-preference shocks. The permanent shocks are real (for example, productivity) shocks and changes in the money supply.4

  • Each of these shocks generates particular movements in the variables of the system that distinguish it from the others (see Panel III). For example, the “real” shock, which is related to factors such as productivity changes or labor market reforms, generates short-run movements (for example, deviations from long-run equilibrium conditions) and trend movements in all variables, whether the latter are real or nominal. The “nominal” shock generates short-run movements in all the variables and trends in only those variables that are nominal. The cost-push shock generates short-run movements in all variables but does not affect their trends.

  • The distinction between these two types of shock constitutes the pillar of the decomposition of each variable into a transitory component and a permanent component (see Panel IV). For each variable, the combination of the short-run movements generated by both types of shock constitutes its transitory component, while that of the effects of the permanent shocks on its trend or long-run dynamics constitutes its permanent component.

The identification of the shocks that drive these two components is based on restrictions that are derived from economic theory. They include restrictions that stem from the long-run equilibrium relationships of a stable money demand, purchasing power parity (PPP), and arbitrage between output and real stock prices.5 The money demand equation embeds the monetarist view that, in the long run, inflation/ deflation is a monetary phenomenon. Purchasing power parity captures the effects of price convergence with trading partners. The arbitrage relationship between output and real stock prices captures the idea that certain developments in the real (supply) side of the economy, such as improved productivity or labor market reforms, can engender trends in asset prices because of their impact on current and prospective levels of corporate profitability. Additional restrictions that include the concept of long-run neutrality of money (vertical Phillips curve) and assumptions on stickiness in the adjustment process of certain variables to shocks are also used to obtain an exact identification of all shocks in the system.

The interpretation of the permanent and transitory components of each variable depends upon the effects of which shocks they include. For example, because the transitory component of output includes the short-run effects of productivity shocks, it cannot be literally considered as a standard measure of output gap that would convey the notion of pressures arising from only the demand side of the economy.

Results

Over the 1998: Q4–2002: Q3 period, the decline in the price level has been associated with a decline in both its transitory and permanent components.6

Although downward pressures on the price level resulting from the decline in the transitory component have been very pronounced during the initial phase of deflation, these have become progressively weaker. Consequently, most of the fall in the price level between 2001: Q3 and 2002: Q3 is accounted for by the decline in its permanent component (Figure 5.2).

Figure 5.2.
Figure 5.2.

Prices: Actual, Permanent, and Transitory Components

Source: IMF staff estimates.

The sustained fall in prices results mostly from the effects of permanent shocks that determine the path of the permanent component of prices and have a substantial impact on the transitory component. Over the deflation period, the effects of transitory shocks on the transitory component of prices have been outweighed by those of permanent shocks (Figure 5.3).7 Shocks such as productivity shocks, changes in the money supply, and price equalization with trading partners have had a significant negative impact on the transitory component of prices over the period 1998: Q4–2002: Q3.

Figure 5.3.
Figure 5.3.

Transitory Component of Prices: Contributions of Transitory Shocks

(In percent)

Source: IMF staff estimates.

In terms of the rate of change in prices (that is, inflation or deflation), the estimates of the permanent component of prices show continued deflation owing to permanent shocks such as productivity shocks, changes in the money supply, and price equalization with trading partners over the last two years (see the graph of the permanent component in Figure 5.4).8

Figure 5.4.
Figure 5.4.

Year-on-Year Inflation Rate: Actual, Permanent, and Transitory Components

(In percent)

Source: IMF staff estimates.

One approach to understanding the relative importance of different shocks is to examine their relative contributions to the variability of prices and output (Table 5.3).

Table 5.3.

Forecast Error Variance Decomposition

article image
Notes: S = short term (1 quarter); M = medium term (12 quarters); and L = long term (40 quarters).

Prices

Permanent shocks contribute 44 percent of the fluctuations in prices over the short term (one quarter) and 94 percent over the long term (40 quarters). The relative contributions of each of these shocks are as follows:

  • Productivity shocks and shocks related to changes in the aggregate money supply and price equalization with trading partners account for 34 percent and 60 percent of the long-term fluctuations in prices, respectively.

  • In the short (one quarter) and medium term (12 quarters), however, productivity shocks are the main sources of variability in prices, accounting for 39 percent and 44 percent of fluctuations in prices, respectively.

The contributions of transitory shocks represent about 56 percent of fluctuations in prices in the short term, which declines to about 35 percent in the medium term. They can be decomposed as follows:

  • Cost-push shocks, which could reflect temporary changes in firms’ markup margins or rates concessions and waivers of water and sewage charges granted by the government, contribute the most to the variability of prices. They explain about 55 percent of fluctuations in prices in the short term and about 13 percent in the medium term.

  • Aggregate-demand shocks, such as discretionary fiscal policies or temporary changes in consumers’ confidence, do not have an immediate effect on prices, but explain about 16 percent of their fluctuations in the medium term.

  • The effects of liquidity-preference shocks on the aggregate level of prices are limited. Liquidity-preference shocks explain only about 6 percent of the fluctuations in prices in the medium term.9

Results (not reported here) of the historical decomposition of the price level into the components attributable to different shocks tell a similar story. Movements in the price level are largely determined by productivity shocks and shocks to the money supply and price equalization with trading partners.

Output

Permanent shocks contribute to 53 percent of output fluctuations in the short term and about 93 percent in the long term, the details of which are as follows:

  • Real shocks (productivity shocks) and nominal shocks (changes in the aggregate money supply and price equalization with trading partners) explain 88 percent and 5 percent of output fluctuations in the long term, respectively. These shocks also account for an important part of its fluctuations in the short term: 44 percent and 9 percent, respectively.

The contributions of the transitory shocks represent 48 percent of output fluctuations in the short term and decline to about 22 percent over the medium term. The respective contributions of each transitory shock are as follows:

  • Cost-push shocks account for 8 percent of the variability of output over the medium term.

  • Aggregate-demand shocks explain about 48 percent of the short-term variability of output. Their relative contribution declines, however, to about 11 percent over the medium term.

  • Liquidity-preference shocks explain only about 3 percent of the variability of output over the medium term.

Another approach to understanding the relative importance of shocks is to analyze their dynamic effects on prices. Such an analysis has several uses. First, it provides a means to verify the consistency of the effects of shocks with standard predictions of economic theory and, therefore, make an assessment of the validity of the identification restrictions used. Second, it provides information about how shocks are propagated and amplified throughout the economy. The dynamic effects on prices of one-time shocks of different types (Figure 5.5) suggest the following:

Figure 5.5.
Figure 5.5.

Movements in Price Level in Response to Different One-Time Shocks

Source: IMF staff estimates.
  • The responses of prices to shocks are consistent with the predictions of standard economic theory. From an aggregate-supply/aggregate-demand perspective, a temporary real shock (for example, a one-time increase in supply owing to higher productivity) that, say, corresponds to an outward shift of the long-run aggregate-supply curve leads to a permanent decrease in prices (and an increase in output). A temporary negative aggregate-demand shock (of a Keynesian style) leads to an inward shift of the aggregate-demand curve that induces a fall in prices in the short run. With downward stickiness in wages, real wages increase, leading to a decline in output and higher unemployment. In the long run, as the aggregate-supply curve flattens, prices go back to their initial levels.10

  • The adjustment of prices to transitory shocks is gradual, which suggests some degree of stickiness. The maximum effect is reached after seven quarters in response to an aggregate-demand shock, four quarters after a cost-push shock, and eight quarters after a liquidity-preference shock.11

Interpreting Results

The large relative contribution of permanent shocks (productivity, money supply/price convergence shocks) to fluctuations in prices, compared with that of temporary shocks such as aggregate-demand and cost-push shocks, probably reflects the increased degree of integration between Hong Kong SAR and the mainland of China and changes in the money supply.12 This implies that downward pressures could continue in the foreseeable future, since (1) price differentials between Hong Kong SAR and mainland cities such as Shenzhen and Guangdong remain substantial; and (2) the stance of monetary policy in the United States could tighten when the economy recovers. Moreover, given that wage differentials between Hong Kong SAR and the mainland cities have not narrowed substantially, the convergence process is very likely to continue in the near future.

The limited contribution and duration of the impact of aggregate-demand shocks on prices, such as temporary fiscal measures, are consistent with evidence on the narrow tax structure and limited size of the fiscal multiplier in Hong Kong SAR. This result implies, however, that macroeconomic policy actions aimed at managing the demand side of the economy—in this case, expansionary fiscal policies—may be unlikely to have a significant direct effect on price developments.

Conclusions

The analysis in this section has shown that the effects of permanent shocks, such as productivity shocks and shocks related to changes in the money supply and price convergence with trading partners, have become more important over time in explaining deflation in Hong Kong SAR. These shocks originate partly from the real side of the economy (for example, changes in productivity) and partly from the monetary side, including the dynamic adjustment of prices for purchasing-power-parity purposes. In addition, the effects of temporary shifts in aggregate demand have been perpetuated by negative wealth and balance-sheet effects in the corporate and household sectors arising from asset-price declines over the past five years. The analysis has also shown that there is a prevalence of productivity and nominal shocks, such as changes in the money supply and price convergence with trading partners, in explaining price and output fluctuations.

1

July 2003 figures indicate that deflation is continuing: prices fell by 4.0 percent year on year, although this sharp decline is partly attributable to temporary utility rate concessions granted by the government.

2

There are three main reasons why stock prices have been used in lieu of property prices. First, the former ensure consistency with the predictions of economic theories that suggest the existence of a stable long-run arbitrage relationship between output and real stock prices (see, for example, Blanchard, 1981). Second, it provides a means for capturing, in a broad sense, the effects of changes in asset prices on both the corporate sector’s balance sheets and households’ wealth. Third, stock prices are highly correlated with property prices—the correlation between stock prices and property prices over 1980: Q4–2002: Q3 is 0.85, suggesting that this approach might not be too restrictive in any case.

3

Note that even the transitory shocks could, through their effects on private sector balance sheets, have persistent effects on prices that last beyond the duration of the shocks themselves.

4

The term “changes in the money supply” refers to increases (decreases) in the money supply beyond (below) what is required to finance long-run real GDP. It is also worth noting that real shocks could also include those changes in the supply of goods and services that are due to wealth/balance-sheet effects resulting from, for example, shifts in investors’ sentiment.

5

These restrictions were tested jointly, since the Johansen maximum-likelihood estimation procedure indicated the existence of three cointegrating relationships. The PPP restriction, when tested separately, was not rejected at the 5 percent significance level. See Becker (1999) and Cassola and Morana (2002) for examples of studies using similar restrictions.

6

The shocks that drive these two components are orthogonal by construction, but there is no restriction that the temporary and permanent components themselves be uncorrelated.

7

The transitory component represents the temporary dynamic effects of all random disturbances and exogenous variables on the variables of the system.

8

The permanent component of the rate of change in prices has been obtained from the estimates of the permanent component of the price level displayed in Figure 5.2.

9

Because these transitory real-asset-price shocks do not create changes in households’ wealth and/or corporate balance sheets that lead to permanent changes in output, they could reflect swings in investors’ sentiment that affect the stock market without affecting the bond market significantly. Such shocks would leave market interest rates unchanged.

10

Although the magnitudes and duration of the transitory effects of these shocks are determined empirically, their zero long-run impact on the price level is imposed by the identification scheme.

11

Although the half-life of deviations from purchasing power parity appears not to be independent of the nature of the shocks that created it, estimates suggest a relatively fast speed of adjustment of the real exchange rate. It takes about eight quarters for half the effects of a cost-push shock to disappear, while half the effects of an aggregate-demand shock disappear only after 12 quarters.

12

Under the linked exchange rate regime, changes in the U.S. federal funds rate lead to comparable changes in Hong Kong SAR’s interest rate (Hong Kong interbank offered rate, or HIBOR). These changes imply adjustments in the monetary base to avoid capital flows that could put pressure on the exchange rate. The relative tightness of the monetary stance in the United States for Hong Kong SAR’s economy can be inferred from the fact that the stock of broad money stood at or below its permanent level—that is, the level of broad money that is required to finance long-run real output—over the period 1996: Q1–2002: Q3.

Meeting the Challenges of Integration with the Mainland