This section discusses a number of issues related to the measurement, determinants, and costs of deflation. It emphasizes that deflation, entailing a sustained decline in aggregate measures of prices, can arise from a variety of sources but that it is seldom benign.
Definition and Recent Experience
Deflation is defined as a sustained decline in an aggregate measure of prices, such as the consumer price index or the GDP deflator. One or two quarters of price declines, while technically constituting deflation, would not be worrying. A mild but continuous deflation could be a cause for concern, however, as it may increase economic uncertainties, distort resource allocation, entail distributional consequences, and lead to subpar growth performance. This definition precludes characterizing deflation in terms of changes in relative prices. In any economy, there will always be sectors where prices are falling in relative, or even absolute, terms. This reflects the normal operation of the market mechanism, whereby changes in sectoral prices play a central role in resource allocation.2
In a number of industrial and emerging market economies, broad indices of goods prices have recently declined (see Figure 1 and Roach, 2002a). But with the weight of goods in CPI falling over time (averaging around only 25 to 35 percent in the Group of Seven economies) and services playing an increasingly important role, it is not clear that deflation in tradable goods sectors per se ought to be a concern. It has been suggested that deflation in goods prices may be a harbinger of developments in the services sector, since many service activities that previously had been essentially domestic oriented or shielded from competitive pressures are no longer so (Roach, 2002b). This has been particularly evident in financial and telecommunications services, but it now applies to several others also. However, there is currently little convincing evidence of a broad-based decline in services prices.
Goods Prices
(Percent change from a year earlier)
Source: Haver Analytics.The impact of large asset price declines on balance sheets and aggregate demand is sometimes associated with deflation. For example, the U.S. stock market crash in 1929 presaged a period of widespread bankruptcies, collapsing demand and prices, and ultimately the Great Depression. The decline in Japanese equity and land prices in the early 1990s has been associated with a prolonged slide in activity and deflation. As Kiyotaki and Moore (1997) show, asset price declines have a severe impact on credit-constrained firms: the declines can spill over from one sector of the economy to another, and their effects can persist over time and become magnified. Nonetheless, while sharp declines in asset prices may reflect a heightened vulnerability, such declines by themselves do not constitute deflation (Bordo, Dueker, and Wheelock, 2001).3
Although the CPI is used as the main indicator of deflation, it is subject to a variety of biases. In almost all countries, consumer price indices are “Laspeyres” base-weighted, and hence do not allow for substitution possibilities. A second source of bias can arise when shifts to lower price discount outlets are inadequately reflected. A third element relates to new or improved products, whereby prices included in the CPI inadequately reflect these products’ characteristics. The magnitude of the overall bias due to these three factors varies across countries, but is generally perceived to range between ½ to 1 percentage point.4
In view of the measurement bias, inflation of less than 1 percent may already be close to price stability, or even reflect declining prices. Table 1 provides an illustration of the incidence of declining prices and inflation below 1 percent for a large sample of industrial and emerging market countries. The estimates show a sharp increase in the proportion of country-months showing annual CPI inflation of less than 1 percent over the past five years. While the higher incidence first became apparent during—and in the aftermath of—the Asian crisis, it has continued over the past three years, especially in emerging markets. A similar picture holds with regard to the more cyclically volatile PPI inflation (Table 2).
Incidence of Deflation and Low Inflation: Consumer Prices1
(In percent)
Number of country months with year-on-year inflation less than 1 percent or negative, as a percent of total. Data based on 35 of the largest industrial and emerging market economies.
Incidence of Deflation and Low Inflation: Consumer Prices1
(In percent)
1980–84 | 1985–90 | 1991–96 | 1997–02 | 1997–99 | 2000–02 | |
---|---|---|---|---|---|---|
Deflation and inflation less than 1 percent | ||||||
All countries | 1.8 | 9.8 | 5.0 | 23.5 | 24.8 | 22.3 |
Industrial countries | 0.1 | 11.8 | 10.2 | 23.7 | 31.0 | 16.3 |
Emerging market countries | 3.2 | 8.3 | 1.4 | 23.4 | 20.7 | 26.3 |
Deflation | ||||||
All countries | 0.9 | 3.2 | 1.2 | 11.4 | 9.7 | 13.1 |
Industrial countries | 0.0 | 3.7 | 2.5 | 7.4 | 6.5 | 8.3 |
Emerging market countries | 1.6 | 2.8 | 0.3 | 14.1 | 11.9 | 16.3 |
Number of country months with year-on-year inflation less than 1 percent or negative, as a percent of total. Data based on 35 of the largest industrial and emerging market economies.
Incidence of Deflation and Low Inflation: Consumer Prices1
(In percent)
1980–84 | 1985–90 | 1991–96 | 1997–02 | 1997–99 | 2000–02 | |
---|---|---|---|---|---|---|
Deflation and inflation less than 1 percent | ||||||
All countries | 1.8 | 9.8 | 5.0 | 23.5 | 24.8 | 22.3 |
Industrial countries | 0.1 | 11.8 | 10.2 | 23.7 | 31.0 | 16.3 |
Emerging market countries | 3.2 | 8.3 | 1.4 | 23.4 | 20.7 | 26.3 |
Deflation | ||||||
All countries | 0.9 | 3.2 | 1.2 | 11.4 | 9.7 | 13.1 |
Industrial countries | 0.0 | 3.7 | 2.5 | 7.4 | 6.5 | 8.3 |
Emerging market countries | 1.6 | 2.8 | 0.3 | 14.1 | 11.9 | 16.3 |
Number of country months with year-on-year inflation less than 1 percent or negative, as a percent of total. Data based on 35 of the largest industrial and emerging market economies.
Incidence of Deflation and Low Inflation: Producer Prices1
(In percent)
Number of country months with year-on-year inflation less than 1 percent or negative, as a percent of total. Data based on 35 of the largest industrial and emerging market economies.
Incidence of Deflation and Low Inflation: Producer Prices1
(In percent)
1980–84 | 1985–90 | 1991–96 | 1997–02 | 1997–99 | 2000–02 | |
---|---|---|---|---|---|---|
Deflation and inflation less than 1 percent | ||||||
All countries | 12.8 | 27.4 | 29.0 | 40.8 | 45.3 | 35.7 |
Industrial countries | 8.3 | 36.0 | 45.5 | 57.2 | 66.2 | 47.3 |
Emerging market countries | 17.2 | 19.6 | 16.5 | 28.7 | 30.2 | 27.0 |
Deflation | ||||||
All countries | 8.0 | 21.7 | 19.1 | 29.1 | 34.0 | 23.6 |
Industrial countries | 2.6 | 27.4 | 30.0 | 37.9 | 45.1 | 29.9 |
Emerging market countries | 13.3 | 16.5 | 10.9 | 22.7 | 26.0 | 18.9 |
Number of country months with year-on-year inflation less than 1 percent or negative, as a percent of total. Data based on 35 of the largest industrial and emerging market economies.
Incidence of Deflation and Low Inflation: Producer Prices1
(In percent)
1980–84 | 1985–90 | 1991–96 | 1997–02 | 1997–99 | 2000–02 | |
---|---|---|---|---|---|---|
Deflation and inflation less than 1 percent | ||||||
All countries | 12.8 | 27.4 | 29.0 | 40.8 | 45.3 | 35.7 |
Industrial countries | 8.3 | 36.0 | 45.5 | 57.2 | 66.2 | 47.3 |
Emerging market countries | 17.2 | 19.6 | 16.5 | 28.7 | 30.2 | 27.0 |
Deflation | ||||||
All countries | 8.0 | 21.7 | 19.1 | 29.1 | 34.0 | 23.6 |
Industrial countries | 2.6 | 27.4 | 30.0 | 37.9 | 45.1 | 29.9 |
Emerging market countries | 13.3 | 16.5 | 10.9 | 22.7 | 26.0 | 18.9 |
Number of country months with year-on-year inflation less than 1 percent or negative, as a percent of total. Data based on 35 of the largest industrial and emerging market economies.
The incidence of an actual decline in prices has also increased markedly in recent years. The number of country months showing a decline in consumer prices rose from around 1 percent in the first half of the last decade, to over 10 percent in the last five years (Table 1). The proportion has continued to increase in the most recent period for both the industrial and emerging market economies. Although less dramatic, the incidence of PPI deflation has increased noticeably, particularly in emerging markets, in the last five years (Table 2).
The background to the increased incidence of deflation is the decline in inflation rates over the past decade (see Appendix Figures A1–A5). Among industrial countries, CPI inflation has declined to an average of below 2 percent, a rate not seen since the 1950s (see Figure 2). Even among emerging market economies, inflation rates are now at their lowest level since detailed data became available in the late 1960s. Low inflation confers significant economic benefits, but it also reduces the buffer against deflation and against hitting the zero bound on nominal interest rates. This raises a number of questions, discussed below, regarding factors that can lead an economy to move from disinflation to deflation, and the costs of deflation.
Determinants
The onset and persistence of deflation can reflect a variety of factors. Consider an economy initially at a full-employment equilibrium (π*, Y*), given by the intersection of aggregate demand (AD) and aggregate supply (AS) (Figure 3). A sufficiently large negative demand shock that shifts the aggregate demand from AD to AD′ can push the economy to a deflationary region (π**, Y**) with declining prices and lower output. Such a shock could reflect a severe cyclical downturn, the bursting of an asset price bubble, or excessively tight policies. The effects of the shock could be amplified through a deterioration in confidence and expectations of declining prices, exacerbating the initial deflationary impulse. The alternative to a demand shock is a positive supply shock that shifts the aggregate supply curve downwards to AS′ where output is higher, even while prices are declining (π′, Y′). Positive supply shocks can arise from a variety of factors, including technological innovation and productivity growth, gains from trade liberalization, or heightened expectations of long-term political and economic stability. An economy may be subject to a combination of demand and supply shocks that could exacerbate the deflationary pressures but ameliorate the effects on output (πc, Yc).
Once prices start declining, the risk of adverse dynamics is heightened, particularly in the case of a demand shock. Following such a shock and the onset of deflation, nominal interest rates will likely decline. Given the zero bound on nominal rates, however, real interest rates will remain positive, and the greater the deflation, the higher the real rates. Monetary policy loses efficacy since demand for money is more or less infinitely elastic at interest rates close to or at zero. In terms of the standard IS-LM framework, this implies that if the IS curve intersects the LM curve at the latter’s flat region, increases in money supply would shift the LM curve to the right to LM′, but have no impact on the interest rate or output (Figure 4). This is the classic liquidity trap.
From an intertemporal perspective, the liquidity trap holds because of entrenched expectations in the presence of nominal rigidities (Krugman, 1998). As long as agents do not believe that the future path of money supply will be changed, they will not respond to current increases in money supply that they deem temporary. This can then become the fundamental challenge to policy in a deflationary environment—to credibly promise to raise the price level not only in the present, but in the future as well (Eggertsson and Woodford, forthcoming).
Expectations are crucial in the persistence of deflation. Suppose that as a result of an adverse demand shock, the price of current goods relative to the price of future goods (P/Pe) is too high, reducing expenditures today in favor of higher (expected) expenditures tomorrow. This means that the economy has an excess of saving over investment in the current period (Krugman, 2002). The equilibrium could be attained by lowering P or raising Pe. However, once the current price level begins to decline, the expected price level may also begin to decline, and activity slows further. As a result, the output-inflation dynamic itself can change. There is a possibility that the Phillips curve flattens, with the result that the aggregate demand shock leads to large declines in output and employment without initially having much impact on prices (Nishizaki and Watanabe, 2000). But pronounced weakness in activity may be followed by an increasing downward pressure on prices. A deflationary spiral can thus form as declining prices reinforce expectations about future price developments.
International Transmission of Deflation
Deflationary impulses can be transmitted across countries. This was not an uncommon occurrence under the gold standard. However, in a global economy with generally flexible exchange rates and independent policy regimes, a generalized international propagation of deflation appears unlikely.
The scope for bilateral transmission of deflation may also be limited. Suppose an economy is witnessing rapid productivity growth and has large excess capacity relative to its trade competitors. Its unit costs of production will be rising at a slower rate than the competitors’ (or even falling), giving it an advantage in the tradable goods sectors. This would lead to an increase in its market share, and, depending on the size of the tradable goods sectors, could result in a large demand shock for the competitors. In response, the competitors may need a depreciation of the real exchange rate, which would normally be achieved by a combination of nominal exchange rate depreciation and lower domestic unit costs (and prices).5 If the nominal exchange rate does not adjust adequately, the burden of adjustment would be on prices. This may cause significant dislocation in these sectors, especially if there are structural rigidities that prevent a reallocation of resources. However, it is unlikely that this process would lead to generalized deflation, given the share of trade (actual and potential) accounted for by any single country, and the possibility of offsetting policy measures. Also, high growth in the large economy could boost demand for imports from the competing economies, partially mitigating the adverse shocks.
Another important impact of an economy undergoing rapid productivity growth—experiencing a decline in unit costs and export prices—will be significant terms of trade gains for its trading partners. This would be reflected in an increase in real incomes and demand growth, which could have substantial positive multiplier effects. These effects could mute the impact of deflationary pressures noted above, depending on the share of tradables in consumption in the importing countries.
Costs of Deflation
Obtaining precise estimates of deflation’s costs is challenging. First, the effects on output of falling prices per se have to be isolated from those of other factors, such as declining equity prices or rising risk premia. Second, it is necessary to quantify the cost of falling prices in terms of output losses. Nonetheless, considerable theoretical and empirical evidence suggest that deflation can be a cause for concern.
Deflation’s costs depend on its source as well as on the extent and duration of falling prices. Deflation may not entail significant costs, and it may be accompanied by growth in output under certain circumstances: these include temporary price declines due to a strong expansion in aggregate supply, as for instance, following a productivity spurt, falling import prices, aggressive trade liberalization, or other positive external shocks. Under such circumstances, deflation is the manifestation of an adjustment to a new equilibrium in the context of rising incomes. It is thus unlikely to become anchored in agents’ expectations and have sustained deleterious effects on domestic demand and the pace of activity. Recently, the information technology revolution, globalization, and deregulation have given a boost to productivity growth across a broad range of goods and services sectors, reducing inflationary pressures. Similar episodes of productivity spurts are believed to have played an important role in deflationary episodes during the nineteenth century (see below).
However, aggregate demand-driven deflation is likely to entail significant costs, in part because of various nominal rigidities. These rigidities can be traced back to financial intermediation, including the zero-interest rate floor, as well as in the labor market. While the estimates due to the former are considered to be significant although difficult to quantify precisely, studies of the downward rigidity of nominal wages suggest that deflation could impose costs amounting to several percentage points of GDP (Akerlof, Dickens, and Perry, 1996).
Collateral and Balance Sheet Effects
In a period of deflation, particularly if unanticipated, the losses of the debtors may not cancel out against the gains of the creditors because collateral loses value. The loss in collateral may be particularly severe when deflation is accompanied by sizable reductions in asset prices.6 The wholesale disposal of collateral could also intensify downward pressure on asset prices. As collateral loses value and bankruptcies ensue, banks need to reevaluate their loan portfolios but may find it hard to distinguish between good and bad risks. As a result, they either raise financing charges (the external finance premium rises) or they cut back on lending that they might otherwise have undertaken, resulting in a fall in aggregate demand.
By reducing collateral from firms’ balance sheets, deflation also reduces the effectiveness of monetary policy in escaping from deflation. Near the zero bound, the policy-determined interest rate plays a less important role in determining the borrowing cost, since the external finance premium becomes increasingly important. In an extreme situation, the economy can end up in a debt-deflation trap. The balance sheet channel also explains why deflation from a positive supply shock may not be as deleterious as from a negative demand shock since rising output and incomes can help offset some of the effects of deflation on balance sheets.
Sticky Wages
Aggregate demand-induced deflation can reduce employment when nominal wages are rigid downwards. With sticky wages, price declines cause real wages to rise, profit margins to fall, and employment to be cut back. This may set off a deflationary cycle. Akerlof, Dickens, and Perry calculate that with 1 percent annual rate of deflation and downward rigidities in nominal wages, unemployment in the United States could rise from a long-run equilibrium rate of 5.8 percent to 10.0 percent. Cumulative output losses could amount to a multiple of the roughly 4 percentage point loss in employment. Most other studies of nominal wage setting, while uncovering significant rigidities, estimate the costs to be less important but still far from insignificant.7
While sticky wages may be destabilizing in response to sustained aggregate demand shocks, they could stabilize output in response to temporary shocks (Keynes, 1936; and Tobin, 1975). DeLong and Summers (1986) show that by inhibiting expectations of sustained price declines from becoming anchored (because consumption is sustained and the likelihood of increases in ex ante real interest rates is reduced), sticky wages may help prevent a deflationary cycle. This may also suggest why structural factors—such as the introduction of a legal minimum wage, unemployment benefits, and support for low income groups—have been associated with a decline in the cyclical volatility of output. However, the failure of wages to adjust in response to sustained declines in demand would eventually lead to rising unemployment and the costs noted above.
Deflation Versus Inflation or Disinflation
Because of the imperfections in financial and labor markets, the costs of deflation and inflation—particularly if unanticipated—are not symmetric, regardless of the underlying economic shocks. The costs in terms of efficiency losses for resource allocation and an increase in uncertainty from moving from zero to very low inflation rates (below 2 percent, say) are likely to be negligible.8 In contrast, the adverse effects due to sustained mild deflation can be substantial.
Deflation and disinflation have some similar effects on activity but, because of market imperfections, deflation entails higher costs. The zero interest rate floor is more likely to become a binding constraint under deflation than disinflation; the same holds for wage rigidities at the zero bound. Furthermore, the “collateral” and “balance sheet” effects may well be stronger under unexpected deflation than disinflation. For example, assessments of creditworthiness typically place a disproportionate weight on the latest loan-to-asset values or loan-to-income/profit ratios. While the fundamental ability of a debtor to service a loan may deteriorate to the same degree under unexpected deflation as under unexpected disinflation, banks are more likely to cut credit lines under deflation for fear of not recovering the book value of the loan principal. Furthermore, the disintermediation that may follow the financial distress triggered by unexpected developments in prices may be larger under deflation than under disinflation. (For historical evidence on the complementarity between price and financial stability, see, for example, Bordo, Dueker, and Wheelock, 2001.)