Milton Friedman famously opined, “Inflation is always and everywhere a monetary phenomenon in the sense that it cannot occur without a more rapid increase in the quantity of money than in output.” However, the equation of exchange, to which Friedman strongly adhered, has three variables (money supply, output, and velocity) that combine to tautologically form price.
Much fat has been chewed over money supply inflation theories, particularly given central bank actions during the financial crisis. This article proposes a framework for understanding inflation with all four terms of the equation of exchange and finds evidence that suggests productive capacity and velocity, not money supply, have the largest impact on prices.
To quickly review the theory: National income can be more than the available money supply because each of us earns money and spends or invests most of it. This spending and investment provides income to other participants in the economy, increasing national income, whereas the quantity of money can remain fixed.
The measure of how many times a dollar circulates is velocity, which means that controlling money supply growth is not adequate to limit demand and thus price growth. A policy that targeted a purely flat money supply would be helpless to contain inflation if the participants were sufficiently motivated to transact with greater circulation of the same dollar. As Figure 1 shows, the velocity of M2 money has varied greatly, ranging from 1.577 to 2.2 over 54 years. The range represents a mammoth $6.74 trillion on the basis of today’s M2 money supply, or 40% of present nominal GDP.
Mervyn King wrote extensively about the relationship between money supply and future prices. The cross-country scattergraphs he generated clearly show a long-run relationship after about the 10-year horizon.
The US data paint a different picture. The monetary aggregate that the Federal Reserve directly controls (monetary base) is actually quite negatively correlated with future inflation. We ran a series of 20 regressions, from 1 to 20 years, comparing the trailing annualized growth in monetary base with the future annualized growth in the CPI (Figure 2).
Out of 20 years, only 2 showed (slightly) positive correlations. If we viewed the relationship in a vacuum, we could perhaps conclude that the Fed’s expanding the base money supply could lower the inflation rate. But that seems logically unlikely, forcing us to search elsewhere for explanations for inflation.
King added a graph comparing money supply and price levels in four instances of hyperinflation: Austria and Hungary in 1921, Argentina in 1989, and Israel in 1995 (Figure 3).
The most fascinating observation of all emerges: In each case, the price level increase preceded the money supply increase — that is, money supply rose to match the increased price level. This explanation is a substantial departure from the typical one, but when a variable lags another, it is unlikely to be a causative factor.
If we cannot explain inflation principally with monetary factors because velocity plays such a critical role, we must find a way to explain inflation with non-monetary factors. To explore this approach, we go back to the equation of exchange, MV = PQ.
To reiterate: M is the money supply (Figure 4), V is the velocity of money, P is the price level, and Q is the output level. The determinants of P and V are where the supply and demand curves meet to form Q. Thus, while Q explicitly is the supply at price P, V opaquely embeds the schedule of demand.
Let’s begin by examining V and, by proxy, demand.
What compels participants in an economy to transact more frequently? One of the more obvious explanations is demographics. In 2011, the San Francisco Federal Reserve published a letter examining the ratio of two demographic cohorts — middle-aged (40s) and old-aged (60s) — against the price-to-earnings ratio, theorizing that the changes drove the marginal preferences in asset markets. Perhaps unsurprisingly, this approach also does a pretty good job of explaining the velocity of money (Figure 5).
But we can do better.
Mimicking the analysis in Figure 2, we ran a series of 20 regressions, from 1 to 20 years, comparing the trailing annualized growth in population with the future annualized growth in the CPI (Figure 6). The R² at 20 years is 0.98 — a very strong relationship that suggests population growth is the primary driver of inflation.
A picture of non-monetary inflation begins to emerge. But we propose that inflation be viewed in the framework of good old-fashioned supply and demand curves (Figure 7). Predict the shapes of those curves and you will predict inflation.
We can begin by taking statistical stock of what correlates best with future inflation. Although we create a simple multiple linear regression model, we logically separate out our independent variables into three categories: supply, demand, and policy (Figure 8).
Output is the product of land, labor, and capital — the primary factors of production. The supply curve is defined by the economy’s productive capacity, which the factors of production form.
We find that the biggest contributor to future inflation at all forecasting intervals is capital stock per capita, which we calculate as the cumulative sum of private investment after capital consumption per person. This supply factor poses as our measurement of the capital of the primary factors, given its convincing correlation (96%) with 10-year future inflation.
Supply is rounded out by the other major factor of production: labor. When the employment-to-population ratio moves up, more labor is available for production, which adds capacity, reducing pressure on price.
Total capacity utilization provides an estimate of how much immediate capacity in all factors of production has been utilized. Unsurprisingly, this factor has the largest correlation at the shortest forecasting interval, with declining significance as the forecasting interval is extended.
As Figure 9 shows, the R² between 20-year (20y) trailing population growth and future nominal GDP growth is 0.935. Future demand is clearly dictated by population growth. A net new person will, at the very least, add net new demand for their subsistence.
But that isn’t to say that the relationship of population to demand is static. When a worker ages into the peak-aged pool, their labor becomes newly valuable and exchangeable. Population growth shifts the demand curve immediately and, over time, across the arc of the expenditures-by-age curve (Figure 10).
The cyclical effects on inflation appear to be very muted other than in the short run. Perhaps surprisingly, the correlation between inflation and unemployment is low or nonexistent at nearly every interval. This result is probably because of the cyclical fluctuations being moderated by policymaking and such automatic stabilizers as unemployment insurance and progressive taxation. Were policy to do less to smooth out downturns in private spending, inflation would likely be more negatively correlated with unemployment. As of now, the strongest correlation is actually a leading correlation — inflation leads unemployment — which is the result of the Fed’s late reaction to inflation, particularly in the 1970s.
Policymaking, fiscal or monetary, really just shapes what trends deliver at the margin. It is less important and less statistically affecting than either supply or demand.
Although M2 is not the strongest input, it does correlate with future inflation, and the growth of M2 is the result of policy over the midterm.
We estimate monetary policy by subtracting the fed funds rate from nominal GDP (Figure 11). This approach is preferred to simple real fed funds because of its superior explanatory power for both future inflation (best R = –0.39 for NGDP – FF versus best R = –0.27 for CPI – FF) and future recessions (best R = –0.29 for GDP – FF versus best R = –0.19).
We estimate the tightness of fiscal policy by measuring the gap between fixed private investment and the deficit net of defense (Figure 12).
We observe that investment is the lever of cyclicality. Since the absolution of Bretton Woods, the deficit net defense has closely offset net private fixed investment. When the investment term becomes too small relative to the deficit term, the economy slows. This result has a much stronger signal for forecasting both the economy and inflation than just the deficit alone because it provides context for how big the deficit should be with respect to the level of net private fixed investment.
The retiring generation of Baby Boomers is offset by the Millennial Generation coming of age. Although the expenditures of retirees decline, they still spend a slim majority of their peak expenditures while not contributing to supply. At the same time, the marginal Millennial coming of age will replace the marginal retiring Boomers’ demand, creating an excess of demand with the same level of labor (Figure 13).
There is substitutability between factors of production. The more capital rich the economy, the less labor intensive is production. This fact is quite meaningful: The dependency ratio in the United States is possibly the best in the Western world, but it is still projected to nearly double in the next four decades. At the same time, dependency ratios in the rest of the world are projected to spike faster and further. This is the biggest story for the supply curve and, consequently, for inflation. The global depletion of peak-aged labor clearly necessitates one of two things: higher prices or higher rates of investment to form new capital.
Although the marginal Millennial coming of age provides an immediate source of labor, turning investment into new capital is not an instantaneous process. In addition, although some demand declines with consumer age and, by extension, demographic composition, residential capital stock is driven more purely by population size.
After being overinvested and hence oversupplied (evaluated with respect to the declining pace of inflation) for a decade, the Great Recession has seen a sharp decline in investment and an even larger decline in per capita investment. The great force that drove down inflation has not only paused but also reversed. In fact, the level of investment at the time of writing is still below replacement (Figure 14).
Inflation and dual-mandate monetary policy truly leave policymakers in a Catch-22. An economy with elevated inflation probably has a real capital deficit. By increasing the cost of capital to reduce demand, formation of fixed capital is also retarded — the deficit of which is the principal determinant of inflation to begin with.
The discussion on inflation has gotten far away from its principal determinants. To return to the Friedman quote, it is easy to conclude that there is some monetary inflation. But inflation is usually and in most places a function of shifting supply and demand curves and not “always and everywhere a monetary phenomenon.” By better understanding the true drivers of inflation, we can forecast it better and manage it more effectively.