By Lawrence H. White
Economists use the term “inflation” to denote an ongoing rise in the general level of prices quoted in units of money. The magnitude of inflation—the inflation rate—is usually reported as the annualized percentage growth of some broad index of money prices. With U.S. dollar prices rising, a one-dollar bill buys less each year. Inflation thus means an ongoing fall in the overall purchasing power of the monetary unit.
Inflation rates vary from year to year and from currency to currency. Since 1950, the U.S. dollar inflation rate, as measured by the December-to-December change in the U.S. Consumer Price Index (CPI), has ranged from a low of −0.7 percent (1954) to a high of 13.3 percent (1979). Since 1991, the rate has stayed between 1.6 percent and 3.3 percent per year. Since 1950 at least eighteen countries have experienced episodes of hyperinflation, in which the CPI inflation rate has soared above 50 percent per month. In recent years, Japan has experienced negative inflation, or “deflation,” of around 1 percent per year, as measured by the Japanese CPI. Central banks in most countries today profess concern with keeping inflation low but positive. Some specify a target range for the inflation rate, typically 1–3 percent.
Although economies on silver and gold standards sometimes experienced inflation, inflation rates in such economies seldom exceeded 2 percent per year, and the overall experience over the centuries was inflation of close to zero. Economies on paper-money standards, which all economies have today, have displayed much more inflation. As Peter Bernholz (2003, p. 1) points out, “the worst excesses of inflation occurred only in the 20th century” in countries where metallic standards were no longer in force. In 1971 the U.S. government cut the U.S. dollar’s last link to gold, ending its commitment to redeem dollars for gold at a fixed rate for foreign central banks. Even among countries that have avoided hyperinflation, inflation rates have generally been higher in the period after 1971. But inflation rates in most countries have been lower since 1985 than they were in 1971–1985.
In the United States, the inflation rate is most commonly measured by the percentage rise in the Consumer Price Index, which is reported monthly by the Bureau of Labor Statistics (BLS). A CPI of 120 in the current period means that it now takes $120 to purchase a representative basket of goods that $100 once purchased. Because the CPI basket is not identical with the specific basket of goods and services that you consume, the percentage rise in the CPI is, at best, only a rough approximation of the percentage rise in your cost of living. The same is true for any alternative measure of inflation, such as the gross domestic product deflator. The GDP deflator is arguably more representative of the economy as a whole, but is less relevant to ordinary consumers because its basket includes the prices of nonconsumer goods (such as new business equipment) that consumers do not buy, and excludes the prices of the many foreign-produced goods that consumers do buy.
Causes of Inflation
In a nutshell, inflation occurs—that is, the purchasing power of the dollar shrinks—to the extent that the nominal supply of dollars grows faster than the real demand to hold dollars. A standard approach to analyzing the connection between the money supply (M) and the general price level (P) uses an accounting identity called the “equation of exchange”:
MV = Py
where V denotes the income-velocity of money (the number of times per year the average dollar turns over in transactions for final goods and services), and y denotes the economy’s real income (as measured, e.g., by real GDP). Because V is defined as Py/M, the ratio of nominal income to money balances, the equation follows. The quantity theory of money (a better name would be “the quantity-of-money theory of the price level”) says that a higher or lower level of M does not cause any permanent change in y or desired V—or, in other words, does not permanently affect the real demand to hold money. It follows that, in the long run, a larger M means a proportionally higher P. In less formal terms, putting more dollars in circulation dilutes the purchasing power of each dollar; or: prices rise when there are more dollars chasing the same amount of goods.
Thought experiments can help to illustrate the thinking behind the quantity theory. Consider an economy in which all prices are in equilibrium. Now, imagine doubling the stock of money by magically doubling the numbers on all pieces of currency and all bank account balances. All price tags must be simultaneously doubled to keep relative prices and the purchasing power of each person’s (nominally doubled) money balances the same, and thus to keep the economy in equilibrium. Prices must rise in proportion to the quantity of money. For a slightly less magical case, imagine that a Federal Reserve helicopter flies across the country and drops enough currency to double the money supply. If the people who get the new cash want to buy the same basket of goods as the population in general, a doubling of all prices is once again called for.
The real-world process by which the Fed injects new money—typically by purchasing bonds in the open market with newly created Fed liabilities—differs from these thought experiments. Among other differences, the first-round spending of the new money is on bonds, not on consumer goods in representative proportions. In the second round, the bond sellers’ banks, into which the Fed has wired newly created reserves, will themselves buy additional securities (or make additional loans), expanding the banking system’s deposits as they do so. The actions of the Fed (and the subsequent actions of the commercial banks) expand the supply of loanable funds and therefore may lower the real interest rate. The commercial banks’ borrowers (predominantly business firms) may, at least temporarily, raise the relative prices of the assets they buy (business plant and equipment). Many economists assume that such relative-price effects are negligible, but others (e.g., the austrian school) assign them a key role in their theories of the business cycle.
For the only result of a real-world monetary expansion to be an exactly equiproportional rise in all prices, the spending diffusion of the new money must not significantly raise some prices ahead of others. This condition is sometimes described by saying that “money is neutral.” In the long run, it is reasonable to assume that relative price effects largely wash out, so for understanding decade-long inflation we may abstract from them. To understand how monetary policy can drive a business cycle, however, the assumption of neutrality must be put aside.
The equation of exchange can be employed to show how the inflation rate depends on the growth rates of M, V, and y. The relationship among all four growth rates is given by the “dynamic,” or growth-rate, version of the equation,
gM + gV = gP + gy,
which says: the rate of growth of the quantity of money, plus the rate of growth of the velocity of money, equals the rate of inflation plus the rate of growth of real income. The equation holds exactly for continuously compounded growth rates. For year-over-year rates it is an approximation.
The dynamic equation of exchange indicates that, as a matter of accounting, inflation depends not only on the rate of monetary expansion, but also on the rate of velocity growth and (negatively) on the rate of real income growth. Which of these three factors contributes the most to inflation in practice? The well-known monetary economist Milton Friedman (1992, p. 262) famously proclaimed: “Inflation is always and everywhere a monetary phenomenon.” What he meant was that sustained inflation has historically always been due to sustained money supply growth, not to sustained velocity growth or sustained negative growth in real income.
The supporting evidence for Friedman’s proposition is straightforward. For virtually any country one examines, even in a bad year real income seldom falls by more than two or three percentage points. Velocity has been known to rise over long periods, but seldom more than one percentage point year after year. When high-inflation and low-inflation countries are compared, differences in money growth are much greater from country to country than differences in either real output growth or velocity. As a result, the rate of monetary expansion is the dominant factor accounting for differences in inflation rates across countries. High-inflation countries are countries with rapid money growth. Likewise, the dominant factor accounting for different inflation rates over decades in the same country (e.g., the lower U.S. inflation rate in the 1990s compared with the 1970s) is different money growth rates. High-inflation decades are decades with rapid money growth. The dominance of money growth in accounting for inflation is especially pronounced in hyperinflation.
The implication for controlling inflation is equally straightforward. Achieving zero inflation merely requires the central bank, which controls the money supply, to refrain from expanding the money supply too rapidly (more specifically, adjusting for velocity growth, expanding the money supply at a rate faster than the economy’s real output of goods and services is expanding). The Federal Reserve System could maintain zero inflation (gP = 0), on average, by controlling growth in the stock of U.S. dollars (gM) appropriately. Central banks elsewhere in the world (Australia, Canada, the euro zone, New Zealand, Sweden, the United Kingdom) have, in recent years, each announced a target range for the inflation rate, often 1–3 percent, and have been rather successful in keeping the inflation rate within that range.
Some economists call the above analysis a “demand-pull” explanation (monetary expansion fuels spending that pulls prices up), while proposing a “cost-push” alternative. For particular episodes of inflation, they have variously blamed monopolies, labor unions, OPEC, and even the failure of the anchovy harvest off Peru for pushing up prices. The equation of exchange warns us that for a “supply shock” to account for a large rise in the general price level (not just a relative rise in some prices, such as the price of oil), the economy’s output must shrink by a large percentage. In practice, “supply shock” cases are seldom large enough to account for much inflation and are typically short-lived. For example, of the 9.2 percent U.S. inflation rate in 1980 (as measured by the GDP deflator, gP = 9.2 percent), the negative growth of real GDP (due, in part, to the OPEC oil price shock of 1979–1980) accounted for only 0.2 percentage points (gy = −0.2%). Meanwhile, growth in the money stock (M1 measure, December 1980 over December 1979) accounted for 7.0 percentage points (gM = 7.0 percent). Growth of approximately 2 percent in the income-velocity of M1 accounted for the remainder (gV = 2.0 percent). For the M2 measure of money and its velocity, the respective figures were 8.5 percent and 0.5 percent.
The equation of exchange also tells us, contrary to what some pundits used to suggest, that “too much growth” cannot be a cause of inflation. The higher the rate of real income growth (gy), the lower the inflation rate (gP), other things (gM and gV) being equal. If an increase in inflation is associated with an “overheating” economy (gy above its sustainable long-run trend), the explanation is that both rising inflation and a temporary spurt in real growth are effects of a previous increase in money growth.
What may look like “cost-push inflation” is often “demand-pull inflation” in disguise. Suppose that expansion of the money supply fuels an increase in demand for retail goods and services. Retailers may delay raising prices of goods already in inventory, but (owing to larger sales) place larger restocking orders with wholesalers, who do likewise, restocking from factories. Factories increase their demands for raw materials and labor, driving up material prices and wage rates. Factories may then “pass through” their cost increases to wholesalers, who do likewise to retailers. At each level, the price increase appears to be pushed by input costs. But the rise in input costs is due, ultimately, to the demand pull of money growth.
Consequences of Inflation
Inflation can do great harm. The harm is greater to the extent that the actual inflation rate differs from the anticipated inflation rate. When transactors correctly anticipate a faster decline in the purchasing power of the dollar (a higher inflation rate), the terms of contracts calling for future payments in dollars are adjusted accordingly. Borrowers and lenders who expect higher inflation agree to a higher nominal interest rate (dollars repaid over dollars lent) so as to preserve the real interest rate (purchasing power repaid over purchasing power lent) between them.
A simple expression for the relation of the nominal interest rate to the expected inflation rate is
(1 + i) = (1 + r) × (1 + gPe),
where i is the nominal rate, r is the real rate, and gPe is the expected inflation rate. This equation is sometimes called the Fisher relationship, after the early-twentieth-century monetary economist irving fisher. Fisher argued that the equilibrium real rate is independent of the expected inflation rate, so that increases in expected inflation are passed through entirely to the nominal rate.
Although lenders and borrowers do not suffer from a higher inflation rate when the rate is perfectly anticipated, holders of non-interest-bearing forms of money, such as currency, do. Higher anticipated inflation subjects them to the equivalent of a higher tax on their money holdings. Inflation thereby drives transactors into costly strategies for getting by with smaller currency holdings, such as making more trips to the bank to take out smaller amounts each time.
From the point of view of eliminating needless costs of economizing on cash, low inflation is clearly preferable to high inflation. But just how low is the best, or optimal, inflation rate? One proposal for achieving the “optimal” result—indeed the most widely discussed proposition in the pure theory of monetary policy—is that the inflation rate should be sufficiently negative that the nominal rate of interest is zero (on bonds of zero default risk and the shortest maturity). Any higher nominal interest rate means that currency pays a poorer return than bonds. This induces people to economize on holding cash, an action that is optimal from the individual’s viewpoint but costly from society’s viewpoint. From the Fisher equation it can be seen that achieving a zero nominal interest rate implies an inflation rate approximately equal to the negative of the real rate of interest, which suggests deflation of about 2–3 percent per annum.
In addition to the tax on cash balances, at least one other harm stems from higher inflation even when perfectly anticipated. With higher inflation, published prices become obsolete more quickly, and so price setters must more frequently incur the costs of adjusting nominal prices. Economists sometimes call these “menu costs” because they include reprinting restaurant menus as well as changing price tags on supermarket shelves, revising catalogs, replacing numbers on gas station price signs, and so on.
Where the tax code is not fully indexed, higher inflation increases the distorting effects of taxes. Before the U.S. income tax brackets were indexed, inflation pushed income earners with unchanged real income into brackets where they faced higher marginal income tax rates. This discouraged people from making taxable income. With indexing of federal tax brackets in 1985, this distortion disappeared. However, the capital gains tax is still levied on nominal gains, not on real—that is, inflation-adjusted—gains. The portion of your asset’s nominal price rise that merely corresponds to inflation is taxed along with any real profit. The higher the inflation rate, the higher the effective tax rate on your real capital gains, even with an unchanged nominal capital gains tax rate. Higher inflation thus discourages capital formation by discouraging people from accumulating taxable assets.
When the inflation rate is incorrectly anticipated, financial trades are upset. If the inflation rate turns out to be higher than anticipated, a borrower gets to repay in less valuable dollars, at the expense of the lender who gets less back in purchasing power than expected. If the inflation rate turns out to be lower than anticipated, the lender gains at the expense of the borrower (assuming the borrower is able to make the greater real payment). For example, the federal government, because it is the U.S. economy’s biggest debtor, gains from unanticipated inflation and loses when inflation is less than anticipated. As a result, the federal government is biased toward higher inflation.
When the future inflation rate is highly uncertain, so that the risk of such gains and losses on new contracts is great, risk-averse parties shy away from making debt contracts (deposits, loans, bonds). Because inflation becomes more variable as the average inflation rate rises, high-inflation economies have stunted banking and bond markets. The real returns from holding bonds and loans of long maturities—for example, thirty-year corporate bonds or thirty-year fixed-rate mortgages—are especially sensitive to inflation variability. When an economy moves to higher and more variable inflation, therefore, such long-term contracts disappear. Long-term investments are discouraged by the greater risk in financing them.
High-inflation currencies have also stunted stock markets, although the reason why is less clear-cut. One likely reason (Aarstol 2000) is that higher inflation is associated with more “noise” in relative prices—that is, transient changes in relative prices due simply to different prices being adjusted at different speeds. Investors, therefore, cannot put as much credence in the earnings reports of companies listed on the stock exchange. High profits for a firm may be only temporary good luck owing to output prices randomly rising ahead of input prices. In such an economy, savers shy away from stock markets, as well as from bond and loan markets. They save less and divert their savings into “inflation hedges” such as houses and gold, rather than adding to the economy’s stock of factories and machines. A second possible reason why inflation reduces the value of corporate shares is that the corporate income tax system in many countries is not fully indexed. Firms face higher real tax burdens as inflation rises.
In addition to hampering financial markets, the “noise” generated by high inflation means that the price system does not communicate information as well (see information and prices). Misinformation distorts investment and employment decisions. For all these reasons, high-inflation economies suffer poor growth. Robert Barro (1997) found in a cross-country study that an inflation rate 10 percentage points higher is associated with real growth 0.3–0.4 percentage points lower. Javier Andrés and Ignacio Hernando (1999), who studied OECD countries, report that lowering inflation by 1 percentage point will boost per capita GDP by 0.5–2.0 percent.
Does inflation have any benefits? Some Keynesian macroeconomists once believed that higher inflation could “buy” a permanent reduction in the unemployment rate, a belief that was encapsulated in early versions of the “Phillips curve.” Economists now agree that no such exploitable trade-off exists; it seemed to exist in the 1960s only when higher inflation was a surprise. Surprise inflation can reduce layoffs (by making dollar sales unexpectedly high) and shorten job search (by making dollar wage offers unexpectedly high), lowering the unemployment rate below its “natural rate.” When workers come to expect a high inflation rate, as they did in the 1970s, unemployment returns to its “natural rate” (see phillips curve). By the same logic, a surprise reduction in inflation can raise unemployment above its natural rate, making disinflation costly.
Although the consensus against high inflation is widespread, opinions vary over whether an inflation rate of 0 percent is better than a rate of +3 percent or −3 percent. There are two main cases in favor of a positive inflation rate. George Akerlof, William Dickens, and George Perry (1996) argue that zero inflation would lead to inefficiency due to wage and price stickiness. In their view, a little bit of inflation provides “grease” to the economic system. Fed governor Ben Bernanke, among others, argues that positive inflation—by keeping nominal interest rates well above their zero lower bound—preserves the Fed’s ability to cut rates if looser monetary policy is needed. In favor of zero inflation, William Poole (1999) finds flaws in both of these arguments. Against the first he notes that if inflation does make nominal wage rigidity easier to live with, for that very reason it likely perpetuates the rigidity. Against the second he counters that low interest rates do not make expansionary monetary policy ineffective. Poole favors zero inflation as the policy that minimizes uncertainty about future inflation, thereby best facilitating financial contracts; and that minimizes the distortions associated with unindexed taxes. In favor of a negative inflation rate there is the “optimum quantity of money” argument noted above for minimizing the deadweight cost of holding currency. On different grounds, George Selgin (1997) makes a case for falling prices in an economy with ongoing productivity improvements. He notes that it is beneficial to let prices of particular products fall as their unit costs fall. In his view, using monetary expansion to raise other prices, so as to produce zero or positive overall inflation, does nothing to increase efficiency but instead increases the adjustment burden placed on the price system.