Why the conventional tools of the Phillips Curve, NAIRU, potential output, and money-supply growth are useless
Cross-posted from INET
The word “unemployment” has a precise technical meaning, with origins in the industrial economy of post-Civil War Massachusetts (Card 2011); to be unemployed is to be seeking paid work but unable to find it at the prevailing wage. The concept was developed for administrative purposes at particular stages of capitalist development; it has legal and social-welfare implications, and the word is not applicable in other settings, such as peasant-agrarian or informal economies.
Though often juxtaposed with unemployment, the word “inflation” has a different ontological foundation. It is a theoretical term that has been bowdlerized in popular discourse, to the point where two distinguished advisers to President Biden recently advanced this prosaic definition: “the rate of change in prices over time.” (Bernstein and Tedeschi 2021). So, it is necessary to distinguish between usages. We might call these “pure inflation” and “everyday inflation.”
2. Pure and everyday inflation
Pure inflation is the theoretical concept. It may be defined as the undifferentiated devaluation of the monetary unit in relation to all goods and services in the economy, on a continuing or sustained basis. This is the type known to acolytes of Milton Friedman as being “always and everywhere a monetary phenomenon.” (Henderson 2021) It is rarely (if ever) encountered in real life. Possibly in 16th-century Europe the influx of silver and gold from the Americas and their effect on the value of metallic monetary units then in use provides an approximate example. The modern hyperinflations and currency collapses of (among others) Germany and Zimbabwe conventionally fall into the same category, even though these undoubtedly had differential effects on exports, imports, and non-tradables. But by contrast, a single once-for-all devaluation (say, Mexico 1995) would not count, if the national money then stabilized, and the price shock passed through the domestic economy within a limited time.
The opposite case, everyday inflation, is of a once-for-all increase in the price of a core commodity – a price shock, typically in energy – that propagates through the general price structure in rough alignment with the factor-intensity of that commodity in different sectors. In the cases of oil and natural gas, direct derivatives such as fertilizer, plastics, and transportation would be hit hard, more remote sectors (such as housing and services) less so. In this case, an increase in the general price level is always observed, because almost all prices of produced goods and services, and especially wages, are sticky downward, so there is never a full offset of increased prices in one sector by decreases in another. However, the net effect is always a shift in the distribution of incomes toward the sectors experiencing the largest price and profit gains, which is why inflation of this type cannot be qualified as “pure.” Further, the shock to the general price level usually dissipates after a certain interval – perhaps normally a few months. It may persist in the data and headlines for longer, as discussed below.
Having identified the two polar cases, “pure” and “everyday” inflation, we may admit the possibility of an intermediate case. This could be called “hybrid” or “persistent everyday” inflation. It would be marked by a sequence of knock-on or ratchet effects (Wood 1978), in which relative price impulses are passed from one sector to another without major damping. A structure of staggered wage contracts across different powerful trade unions could have this quality, with wage and then price increases ricocheting from one industrial sector or public service to the next. The US and UK inflations of the 1950s through the 1970s were more-than-possibly of this type.
With this typology in mind, the US price increases of 2021-2022 were certainly an everyday inflation. There was no collapse of the US dollar on international markets, nor any general, undifferentiated increase of all prices. There was also no long-term reverberation of cost pressures from one sector to the next, and no evidence of an ongoing wage-wage spiral. What did happen, was a series of cost-shocks related to the pandemic and its aftermath, briefly exacerbated by the sudden escalation in early 2022 of the Ukraine war, which began to fade from the data over the second half of 2022.
3. Sources of everyday inflation in 2021 and 2022
The most important cost shock was in the energy sector, and specifically oil. With the onset of the pandemic, sales, production, exploration, and pricing in the domestic US oil sector all declined sharply. This created a low base for the recovery of oil prices in 2021 and into 2022, hence a large hit to the rate of change when recovery occurred. The shock also meant that oil properties, notably in the Permian Basin, were temporarily cheap. Private equity moved in, stating openly to the local press that their objective would be to increase the rate-of-return and “shareholder value” – rather than maximum levels of production or economic growth. A typical report in the Houston Chronicle reads:
Investors in oil and gas companies, however, have been pushing for ‘capital discipline’ and increased returns. The result is, instead of spending to quickly ramp up production in oil fields such as the Permian, companies are sticking to already planned production increases — providing only modest relief to tight supplies and high prices while passing on a good chunk of their blockbuster profits to investors. (Buckley 2022)
The price of oil had fluctuated in a range roughly from $65 to $80 per barrel (WTI, adjusted for inflation) in the years just before the pandemic. It took a spectacular dive to just $20/bbl in early 2020, recovered to its pre-pandemic levels, and then briefly spiked to around $116/bbl in early 2022 before again returning to $80.55 in November 2022. In inflation-adjusted terms, the price of oil never reached levels prevailing as recently as 2014 (FRED 2022), yet from the low base of early 2020 the rate of change, and therefore the contribution of derivative fuels to the change in the consumer price index, was dramatic. It was, however, finished by June 2022, with price deflation setting in thereafter. In the interim, oil prices drove the gasoline component of the Consumer Price Index up by 154 percent from the low in March 2020 to the peak in June 2022, with indirect effects on food and all other sectors.
A second cost shock affected automobiles. Here the culprit was a shortage of semiconductors, necessary for new cars. In the early days of the pandemic, with a sharp decline in commuting, major semiconductor producers bet on a shift of demand toward electronic appliances and household equipment, which did not occur. New car production was therefore well short of demand as 2021 came around. The effect on new car prices was modest, as the main effect in that market was backlogs and queues. However, demand was displaced onto used cars, which exist in fixed supply and sell for what the market will bear. Used car prices rose 55 percent to a peak in February 2022.
A third significant (though smaller, and considerably later) price increase occurred in the housing component of the CPI, which accounts for about thirty percent of the index. In this component, actual rents largely stand in for the cost of housing; the idea is that a shift from rental to ownership of a house should not affect its contribution to measured output. In practice, the imputation is problematic. Rental markets are lower income, lower quality, and higher turnover than sale markets, and the price of new rentals is more volatile than rents under long-term contracts, which are in turn more volatile than the actual cost of maintaining a home owned outright or with a fixed mortgage.
Thus, it is possible that an increasing price of new rental contracts may have an amplified effect on “imputed rents” — which homeowners are calculated as paying to themselves — while having little material effect on the actual housing costs of most American homeowners. The housing component of the price index accelerated from the summer of 2021 through the fall of 2022, at which point it too began to subside. In any event, as with used cars, the sale or rental of existing homes is an internal transfer, with equal gains and losses on either side of the transaction. It is not a relationship between “consumers” and “producers” of a good or service. It is not clear why this component should figure heavily, or at all, in policy decisions over “inflation.”
The issue of persistence is blurred by transmission effects from the wholesale to the retail levels. While oil prices had recovered just to previously normal values by March 2021 (and only thereafter spiked briefly in March 2022), gasoline prices, which figure directly in the CPI, continued to rise until a peak in June 2022. Is this “persistence” – or is it merely a lagged effect, as retailers sell off inventories acquired at lower prices and replace them with new products at higher prices?
For the overall price level, there was a further, fateful, illusion of persistence. The Bureau of Labor Statistics reports the change in the CPI on a 12-month basis, evidently to avoid the flux endemic to a monthly survey. This practice has the consequence of generating eleven additional headlines after any one-month shock to the price level, each of which may contain no new information whatever. Such news stories (and associated opinion pieces) continued through the mid-term elections of 2022. By then, repetition had done its job; the course of policy was set. Afterward, it became clear that there was no persistence of price pressures in the US economy (Smith and Duguid 2022), and that an evident turning point had already been reached by mid-summer.
4. The Phillips curve, the NAIRU, and inflation targeting
Prominent economists – Lawrence Summers (2021), Jason Furman (2022) – were quick to fix the blame for the rising prices of 2021 and 2022 on macroeconomic policy, and specifically fiscal policy, while others – Kenneth Rogoff (2022), Alan Blinder (2022) – placed the spotlight (if not the blame) on the allegedly over-expansionary (or insufficiently reactive) policies of the Federal Reserve. Although their interpretations differ in some respects, these views are all rooted in the history, as they saw it, of the inflation of the 1970s, and in the models common to that era.
We may distinguish perhaps three variants of this macroeconomic view. The first adverts to the Phillips Curve (Samuelson and Solow 1960), which hypothesized, from limited observation and bold conjecture, a stable tradeoff between unemployment and inflation rates. The Phillips conjecture gained credibility from rising prices and falling jobless rates through the 1960s in the United States, but failed to correspond to the facts thereafter, and never did in most other industrial countries.
A second variant emerged at the hands of Milton Friedman (1968) and Edmund Phelps (1967); they introduced a hypothetical called “inflation expectations” and showed that incorporating that concept into a Phillips equation would generate a near-vertical long-run relationship between inflation and unemployment. Thus emerged the “natural rate” or “non-accelerating inflation rate” of unemployment (NAIRU). By the lights of this model, attempts to reduce unemployment (below its “natural” value) with demand stimulus could lead to ever-accelerating inflation. In a theoretically-consequential but empirically-nonexistent gloss, Robert Lucas (1972) and others modified the expectations idea (from “adaptive” to “rational”) to produce a strictly-vertical inflation/unemployment relationship and the “hyper-neutrality” of money. By dint of logic and in the face of the collapse of the original Samuelson-Solow formulation, Friedman-Phelps-Lucas underpinned a new orthodoxy, according to which central banks should only target the inflation rate and leave unemployment to the “labor market.”
Exactly how the central bank should target the inflation rate is an awkward question to which no clear answer exists. The original counter-Keynesian (monetarist) view held that the central bank could and should target the growth of the money supply. Application of this idea from 1979 to 1982 came with the cost of a massive slump and global debt crisis; in the aftermath, the practice of money growth targets was abandoned. Since then, Federal Reserve policy combines targeting of short-term interest rates with public statements on desired inflation; the effectiveness of this strategy was never tested in the forty years after Volcker dumped monetarism, as there was no inflation to test it on. Nor (despite residual monetarist fears) did “quantitative easing” in the 2000s bring inflation back. As Hyman Minsky observed (Marselli 1993), banks do not lend reserves and they do not need reserves in order to lend.
By 2018, the evidence against the natural rate/NAIRU hypothesis was strong enough for Olivier Blanchard (2018) to raise some tentative questions and to suggest that economists might “keep an open mind and put some weight on the alternatives.” There is no sign that this wise advice took hold.
5. The output-gap model
With the demise of the original Phillips Curve, some mainstream economists who were unwilling to give up a stabilizing role for macroeconomic policy retreated to a rough-and-ready calculation of “economic slack” – the estimated difference between actual output and its “potential” value. The latter could be calculated by projecting forward, from the preceding peak, the growth rate of real Gross Domestic Product before the downturn. The underlying thinking was that so long as fiscal stimulus was limited to an amount estimated to be sufficient to close the actual-to-potential output gap, inflation risks were minor and could be disregarded. This view became highly influential in setting fiscal policy.
The output-gap model lends support to the leading pseudo-Keynesian prescription in the face of recessions, namely tax cuts for households and business firms combined with direct cash transfers to households. These measures, universally described as “stimulus,” were thought to be the fast-acting solutions; fresh money in private hands would be quickly spent, bringing on the most rapid recovery and return to high employment. Direct spending – public works and jobs programs – would operate directly on GDP (as tax cuts and bonus checks do not) and with larger multiplier effects. But – the argument generally went – they could not be mobilized so quickly, when it mattered most. They would thus risk coming onstream just as the economic recovery was complete, and so adding, counterproductively, to renewed inflation.
While the output-gap approach was (and to many economists still is) intuitively appealing, it rests on several unstated assumptions. The first is that slumps are mainly driven by cycles in effective demand, whether for consumption goods, business investment, or exports – and not by changes in the material conditions underpinning production. A decline in resource quality, an increase in resource costs, institutional failures, or a technological shift rendering accumulated physical capital obsolete – these matters would preclude any early return to the previous production peaks, and render calculations based on those peaks irrelevant to the post-slump conditions. Second, even if the calculation is correct when first made, any failure to return rapidly to the previous peak production entails the decay of existing capital, so that the potential dips toward the actual as time passes. If these possibilities are acknowledged, the output-gap calculation would appear to leave even less room for a non-inflationary pseudo-Keynesian stimulus policy.
Further, the amount that seemed desirable would depend also on the rate at which growth was forecast to resume in the absence of a stimulus policy. Since standard models build in a reversion to the previous equilibrium path, quoi que vienne, an optimistic forecast militates for cautious estimates of how much “stimulus” to enact. In 2009, caution prevailed, with a “targeted and temporary” expansion program including a large tax-cut component. It proved seriously insufficient. In 2020 and 2021, very much larger and more sweeping emergency programs evoked expressions of concern from leading figures of the Obama era; Summers (2021) in particular warned that it could set off an accelerating inflation as households rushed to spend their income gains.
There is, however, a factor that cuts in the opposite direction – as indeed Milton Friedman (1957) once implied, in advancing the “permanent income hypothesis” for which he was awarded the economists’ Nobel prize. Friedman argued that, past a certain point, a large transfer of purchasing power to households may affect balance sheets rather than behavior. In that case, the consequences of fiscal stimulus would be largely limited to the preservation of previous spending patterns; any additional transfers would merely be stored away, partly as a hedge against future ill fortune. Curiously, though he had expressed admiration for Friedman on points on which he was wrong (Summers 2006), Summers did not (so far as I’ve seen) refer to the permanent income hypothesis in the debates of 2021, although it proved in this instance to be a decent guide to actual events.
6. US household behavior in slump and stimulus
The United States had a poor and badly disorganized public health response to the onset of the Covid-19 pandemic, with a rapid spread of the virus and a chilling death toll, albeit heavily weighted toward the elderly, immunocompromised, and otherwise at-risk parts of the population. However, the economic response was rapid and administratively efficient, providing a cash transfer through the tax system and a very substantial extended unemployment benefit, which, at $600 per week initially, represented a raise for a large segment of the working population. For this reason, despite a spectacular collapse in employment and working incomes, there was little-to-no increase in poverty and certain indicators of well-being, such as food insecurity, actually declined (Schanzenbach 2022).
What did households do with the money? For the vast majority, laid-off from low-to-moderate-wage service jobs, the best answer (Peterson Foundation 2021) appears to be that they kept up with their ordinary, fixed, and customary expenses: rent, utilities, groceries, gasoline, education. They did not splurge, but for these households, there was only a modest increase in savings. For households in the upper tiers of the distribution, the picture is different. They had previously spent heavily on the services that employed the large majority of American workers; they were less likely to become unemployed and the assistance was a less-important share of their incomes. But they were cut off from the ordinary use of their earnings. So they saved what they could not spend, and aggregate savings rose temporarily to about one-third of aggregate income. These savings then found their way into asset markets: real estate, corporate equities, collectibles, and the like, with purchases abetted by extremely low long-term interest rates. Asset prices, accordingly, recovered quickly and rose sharply as the pandemic wore on.
None of this supports the notion of an inflationary spending spree fed by a reckless “stimulus” policy. There was, to a degree, a shift of purchasing power, blocked from services, into household appliances, vehicles, home renovations, and new construction. But (as institutionalist theory would predict) the result of this shift toward newly-produced goods was (mainly) backlogs and queues and shortages rather than price increases, and in many cases the backlogs were of imports, leading to epic congestion in the container ports of the US West Coast (USDOT 2022). Price effects were (again, as theory would expect) stronger in asset markets. As stated above, these are not markets for produced goods and therefore not normally considered to be elements of inflation, even if they do appear in some components of the Consumer Price Index. On the contrary, the usual word for a general increase in asset prices is “boom.” And booms, as history shows, are deeply vulnerable to increased interest rates.
In brief summary: The price increases of 2021-2022 were cost-driven, accompanied by an asset price boom incident to the disruption of the service economy. They were not driven by macroeconomic excess, neither fiscal nor monetary. But they did hand the Federal Reserve a political problem, which it proceeded to solve, in what may prove to be the worst way.
7. The Fed waves its wand
According to President Biden (Irwin 2022) and to the large consensus of mainstream economists and the voices of the financial sector, the “inflation” problem of 2021-2022 fell within the responsibility of the Federal Reserve. This was convenient for each player in the drama. For the President, it meant that political responsibility for price increases and (worse) for the consequences of dealing with it could be shuffled off onto an impregnable institution outside his control. For the banks, vested through quantitative easing with vast excess reserves, it would mean earnings without risk or effort, since the Federal Reserve pays interest on reserves at the official rate. For the economists and central bankers, it would mean vindication of their long-held beliefs and a boost to their perceived influence. The costs would fall elsewhere – on other countries and their banks, on speculative markets, on homebuilders and homeowners, on the indebted, and, eventually although not necessarily soon, on businesses and the presently employed.
The Federal Reserve therefore acted. Interest rates rose in large tranches from early 2022 through the late fall. Higher interest rates quickly quelled the housing market, while supporting the dollar and therefore keeping a lid on the price of imports. Stocks, especially in the technology sector, and cryptocurrencies fell in value. By raising interest rates aggressively, the Federal Reserve also received credit for an end to price increases in core commodities that would have stopped rising in any event, especially after the administration started selling petroleum from the Strategic Reserve and oil prices were brought back down. The fact that absolutely no prior theory or evidence supports the notion that tight monetary policies can end inflation within just a few months was, in the main, conveniently overlooked (Galbraith 2022b). The Federal Reserve is a very lucky institution.
However, there is a fly in the soup. It is the relationship between the short-term interest rate, which the Federal Reserve controls, and the longer-term rates, on Treasury bonds and in the private sector, over which the central bank exercises very little immediate influence. Long-term rates, for a safe asset like Treasury bonds, are a compound of current short-term rates and the expected future course of short-term rates over the lifetime of the bond. This second element has been conditioned (very reasonably) for years to expect very low short-term rates, and thus to view a rise in rates as a temporary aberration, likely to be reversed once the economy falls into a deep enough slump. The result is that the yield curve, normally upward-sloping, is now inverted. There is therefore no reason for any investor to buy or hold a long-term security – the short-term assets are not only safer, but also a better deal.
This is why an inverted yield curve is almost always followed by a slump in business investment, home construction, housing prices (and therefore the viability of mortgages), and of course in stocks and bond markets (Galbraith, Giovannoni and Russo 2007). At present writing, many observers can see the coming storm. But the Federal Reserve is stuck. If it relaxes policy, it will appear over-sensitive to economic risks, inconstant, and non-credible. If it continues down the present path, it is steering economic activity toward a cliff. Again, not necessarily soon. But in due course, and inevitably – opening the door to another financial crisis and yet another round of crisis interventions.
In this climate, at present writing, the Federal Reserve’s leadership has shown its colors and commitments (Galbraith 2022a). The interests of the dollar and the US banking sector in world competition are paramount. So far as the domestic political economy is concerned, the priority is primarily toward ensuring that wages never catch up to the price increases that have already occurred. In recent speeches (Cohan 2022), Chairman Powell has made this commitment abundantly clear – and even the fact that wage growth has slowed in recent months seems to be making little impression on the course of policy. Unemployment must rise, labor markets must soften, capital must gain, and workers must lose. That is where matters presently stand.
8. The prospect for prices
The most likely course of events is therefore a renewed slump and a further shock to employment and wages. Apart from construction, however, this may not happen soon; contrary to commonly expressed liberal worries (Olander, 2022), the next big one may not yet be in sight. Businesses and households have a way of trying to survive, when conditions begin to get worse, by taking on new debts even on unfavorable terms. Lenders usually find such deals attractive; they bring good returns and the assets, such as they are, can often be securitized and fobbed off on the unwary. This can continue until it stops.
And so, it also remains possible that the crisis of prices, such as it was, may not have run its course. There are three reasons to fear additional problems on the cost front in the period ahead.
First, there is a problem of gross markups. In normal times, with general price stability, these are stabilized by convention and habit, by the economic equivalent of good manners. Businesses are cautious about antagonizing their customers; they do not like to acquire the reputation of a price gouger. (This is why, for instance, hardware stores do not generally jack up the price of plywood when a hurricane is on the way). But in a general melee, with prices going up all around, a different mentality sets in, an impulse to grab what one can, and not be the sucker left behind. An inflation driven by profits (Bivens 2022), not wages, can therefore reverberate for some time. Unlike a wage spiral, a profit spiral gets little media attention and policy response – for obvious reasons.
Second, there is a risk of more shocks to core commodities, especially in the energy sector. Oil prices came down in 2022 thanks to sales from a finite strategic reserve. Now with the mid-term election past, the administration plans to buy oil from the market to replenish the stock (White House 2022). Will production suffice to cover both regular demand and storage? So far as known, no one really knows; both the geology and the strategy of the producing firms and refiners are uncertain. But energy markets are financialized, and there is in them the capacity for speculative manipulation – what I have called the choke-chain effect (Galbraith 2014). We shall see whether we are in for another round of that.
Third, there is the effect of higher interest rates on business costs. Interest, after all, must be paid. Sooner or later, the higher short-term rates will bleed into the accounts of business borrowers, and some of the effect will be passed along, so far as conditions permit, to their customers. To that degree, a tight monetary policy is inflationary before it is disinflationary.
In sum, the theoretical construct of pure inflation is of no use in understanding the price events of 2021 and 2022 in the United States. By extension, the conventional tools of the Phillips Curve, NAIRU, potential output, and money-supply growth are equally useless. By further extension, the “anti-inflation” policies of the Federal Reserve have acted on asset markets (which are not part of theoretical inflation) while taking credit for the end to a price process in produced goods that was transitory in any event. Yet the Federal Reserve is now stuck in a posture guaranteed to destabilize economic activity sooner or later, while the economy remains vulnerable to additional potential price shocks emanating from the same sources already seen, including real resources, supply chains, wars, pandemics, and the policies of the Federal Reserve itself. These can be dealt with, if at all, only by policies in each specific area (Weber 2021). And that, by the way, was very much part of the thinking of John Maynard Keynes (1940) on this topic in How to Pay for the War, along with the practice of his partial disciple, John Kenneth Galbraith (1952), at the Office of Price Administration in 1942-1943.