What this covers

Robert Hall of Stanford, speaking with EconTalk host Russ Roberts, challenges the narrative of a sluggish recovery from the Great Recession. Hall argues that the labor market actually recovered along its normal historical path—unemployment sits well below the long-run average and job-filling is at record difficulty, both signs of a tight market. The disappointment people feel is not a symptom of incomplete recession recovery but rather of two separate and persisting problems: slow productivity growth and weak capital formation since 2007. The conversation centers on Federal Reserve mechanics and the proper tools for monetary policy, drawing on Hall's experience chairing the National Bureau of Economic Research Committee on Business Cycle Dating.

The discussion spans the architecture of the banking system, the long-term trajectory of labor force participation, and the nature of recent productivity slowdowns. Hall contends that the banking system functions as a sealed system, meaning banks collectively cannot choose their reserve levels—only the Federal Reserve does, via interest on reserves. He recasts how monetary policy works in a regime where reserves earn interest: Milton Friedman's hot potato effect, where banks tried to lend out reserves, now operates through the interest rate itself rather than reserve quantity. The conversation also takes up questions of measurement (why traditional GDP definitions of recession break down), market concentration as a drag on investment, and the limits of policy uncertainty as an explanation for stagnation. Hall emphasizes productivity uncertainty rather than pessimism, noting that six-year slowdowns are statistically routine and have typically reversed.

Sharpest takeaway

Hall argues the post-2008 'slow recovery' is a misnomer: the labor market recovered along normal historical lines, and the lingering disappointment comes from slow productivity growth and weak capital formation, not an incomplete recovery; meanwhile, the Fed's reserve mechanics mean banks collectively cannot reduce reserves and the interest rate on reserves is the proper policy instrument.

  • The labor market behaved as it historically does after a large shock and is back to normal conditions
  • Banks collectively have no choice over reserve levels; only the Fed sets them via the interest rate on reserves
  • Productivity slowdowns of several years are statistically common and typically reverse

The argument · threads7 threads · 27 claims
0.80

Productivity growth slowed markedly after 2007 and remains the primary driver of disappointing output, distinct from capital formation effects.

5 pointscentrality 5/5
  • Low capital formation and low productivity are not the same thing: total factor productivity is the residual by which output grows more than expected given rising labor and capital, so it already accounts for the fact that capital formation was weak, meaning the lost capital that would have formed without the recession is a separate, nontrivial drag on output.

    “the residual is also called 'total factor productivity'. The 'total factor' means that it's considering and giving credit to all factors, capital and labor, and not just labor.”

  • Claims that all the good ideas have been implemented are a tired, recurring story—productivity pessimism around 1990 preceded the internet boom—and slowdowns over six-year periods are statistically common and have typically been followed by productivity growth returning to normal or above normal, so the right framing is productivity uncertainty, not pessimism.

    “Episodes like we've had recently, of a slowdown in, say, over a 6-year period, have been common. There's nothing statistically surprising about it. And they've typically been followed by productivity growth going to normal or even above normal.”

  • Computer vision illustrates ongoing productivity potential: trash collection went from a three-person crew to one driver (a threefold labor reduction) while operating at least three times faster, yielding roughly a ninefold increase in productivity, and this only scratches the surface of computer vision applications being pursued by many tech startups.

    “they have a 3-fold reduction in the number of people, from 3 to 1. And you have a tripling of the speed. So, the productivity of collecting trash has gone up by a factor of 9.”

  • While the economy genuinely produces unmeasured benefits (e.g., the photography revolution is not captured in data still based on film), Chad Syverson's work shows the upper limit on such mismeasured gains is fairly limited, because most of what the economy produces—electricity generation, home building—uses essentially unchanged technology and represents large budget shares with little productivity growth.

    “Chad Syverson, Chicago, has written a very interesting paper... that upper limit is pretty limited. The idea that we're somehow missing some huge benefits.”

  • The most important factor in the disappointing recovery is the slowdown in productivity growth—productivity still grew but at much lower rates—and this is fundamental to the amount of output, more important than the unexpected several-percentage-point drop in labor force participation.

    “Even more important is simply the slowdown in productivity growth. We had productivity growth, but at much lower rates. And that's just fundamental to the amount of output.”

0.80

Monetary policy operated near its limits; the Fed's proper instrument is now the interest rate on reserves, which was kept too high.

5 pointscentrality 5/5
  • With $3 trillion of reserves outstanding (versus $17 billion historically), the Fed must keep the rate it pays on reserves close to market rates because identical components of the federal debt cannot durably pay different interest rates without creating disequilibrium, so the Fed will track rising market rates upward rather than ever allow a large differential.

    “with $3 trillion dollars of reserves outstanding, they have to make those attractive enough... we have to keep that differential fairly small because we've got $3 trillion dollars of reserves out there.”

  • The traditional Friedman quantity-of-reserves theory—where boosting reserves makes them hot potatoes banks try to lend out, expanding the economy until a new equilibrium—was built when reserves earned zero interest; in the modern regime the equivalent lever is the interest rate paid on reserves, where lowering it makes reserves hot potatoes and expands the economy.

    “the corresponding thing to increasing the quantity of reserves is to lower the interest rate; that low interest rate makes reserves hot potatoes; that expands the economy.”

  • The interest rate the Fed pays on reserves should be the primary instrument of monetary policy, and the rate paid has been too high—it arguably should have been negative (the ECB paid -40 basis points), and the Fed's 50 basis points is well above market, a subsidy that at one point flowed to foreign banks like Deutsche Bank as the largest holder of excess reserves harvesting money from American taxpayers.

    “my feeling is that it should be the instrument of monetary policy. But if you asked then what I thought the rate should be, I would have said it should be negative.”

  • Quantitative easing is most instructively understood as the Fed borrowing funds from commercial banks—by creating reserves, which are Federal obligations—and using the proceeds to buy bonds; it greatly expanded the national debt and is mechanically just a bank getting funds from one source to buy a different type of security, not electronic magic.

    “They borrowed the funds from the commercial banks and used the proceeds to buy bonds... The Fed greatly expanded the national debt by issuing what are Federal obligations, borrowing--which are reserves. It's as simple as that.”

  • Concerns about an inflationary unwind are unfounded because the Fed has two equally viable normalization strategies—either keep the reserve rate low and shrink the portfolio back toward $17 billion, or pay market rates on reserves and keep trillions on the balance sheet—so the large balance sheet is just a different form of government debt, not inherently inflationary.

    “We could either keep the rate we pay on reserves at a low level and reduce our sales back to $17 billion of reserves... Or, Strategy Two is we can pay market rates on reserves and we can continue to have trillions. And either one of those works fine... belt and suspenders.”

0.61

Discretionary fiscal stimulus failed because funds went to debt and tax cuts rather than government purchases, teaching nothing about multipliers.

4 pointscentrality 4/5
  • There is a long literature showing that discretionary policy aimed at increasing government purchases like infrastructure doesn't work, or works perversely: it takes so long to crank up spending that the stimulus arrives only when the economy is already back in a boom and no longer needs it.

    “discretionary policy aimed at increasing government purchases like infrastructure just don't work. Or, if they do work, they don't start stimulating the economy until it no longer needs it... it's when the economy is back into a boom condition.”

  • Because the stimulus did not actually translate into increased government purchases—the funds went to debt paydown and tax rebates—it teaches nothing about the multiplier; you can only learn about the multiplier when government purchases actually change, which is exactly what failed to happen over the last decade.

    “you learn nothing about the multiplier if nothing happens to government purchases. Which is exactly what happened over the last decade.”

  • There is broad consensus that government purchases raise GDP: away from the zero lower bound, a billion dollars of government purchases adds about a billion dollars to GDP (a multiplier of about 1, even assuming monetary offset), and at the zero lower bound the multiplier is higher; a multiplier of 0 is outside the range of nearly all empirical work and modeling—including Ramey's and Barro's, which both find positive multipliers.

    “a billion dollars of government purchases adds about a billion dollars to GDP. So, the so-called multiplier is 1... A multiplier of 0, which would mean no effect, is outside the range of almost all empirical work and almost all modeling. It would not be a viable position.”

  • Calling the stimulus an $800 billion injection is misleading because a large share went to state and local governments who used it to pay off debt rather than expand spending, so the National Income Accounts show no expansion in state and local government purchases—in fact a contraction.

    “a large amount of that money went to state and local governments. And rather than expanding their fiscal effects, a lot of them... used them to pay off debt... you don't see any sign whatsoever. In fact, you see a contraction.”

0.56

The labor market recovered normally after a massive shock; lingering weakness stems from benefit traps and long-term participation decline, not recession damage.

3 pointscentrality 4/5
  • The labor market is back to absolutely normal conditions: unemployment at 4.6% is well below the long-run average of 5.9%, and it takes longer than ever before to fill a typical job, which is a sign of a tight labor market—so there is no lingering recession effect in the labor market.

    “The unemployment rate of 4.6% is well below the long-run average of 5.9% for unemployment. Key factors like how long it takes an employer to fill a job--it takes longer than it ever has before to fill a typical job. That's a sign of a tight labor market.”

  • The decline in prime-age male labor force participation is a very long-term phenomenon stretching back to 1950 along a smooth downward trend, contrasting with women, whose participation rose then began declining 10-15 years ago; teenage employment has roughly halved from about half of teenagers working in 2000 to under 25% today, tied to declining health across age groups and absorption into Disability.

    “in the year 2000, not that long ago, about half of all teenagers at any given time were working. Today, less than 25%. So there's been a decline in half of the fraction of teenagers who are employed.”

  • The system has created programs where people who enter the labor market lose benefits so aggressively that they become trapped, making it irrational to take a job even when the labor market is normal; benefit reform—especially Disability—is acutely needed, though the claim that the entire recession was caused by benefit expansion is wrong because the 8 million who lost jobs in 2008-09 demonstrably did not quit.

    “we've created a system where people become dependent on programs, lose their benefits so aggressively that they become trapped. It just doesn't make sense to enter the labor market even though it's back to normal because of losing benefits.”

0.38

Market power concentration and policy uncertainty function as drags on investment, but uncertainty's causal effect on spending has proven small empirically.

2 pointscentrality 3/5
  • An increase in market power and concentration in the economy functions like a tax that discourages economic activity, including capital formation and possibly productivity, making it a more promising explanation for stagnation than policy uncertainty—though market power remains remarkably elusive to measure.

    “Market power functions like a tax to discourage economic activity... there has been some increase in concentration... I think this market power hypothesis is going to get a lot of attention as another thing that's holding back, and maybe even holding back productivity.”

  • Although policy uncertainty is high, establishing a direct causal connection between uncertainty and reduced investment has proven elusive in the literature; 9/11 served as a natural laboratory—uncertainty spiked but the dip in investment and consumption was brief and the economy resumed normal behavior within months—showing the effect is real but remarkably small.

    “getting a direct sort of causal connection has proven to be elusive... 9/11... There was a brief period of diminished spending... And then... people looked around and said, 'I guess not.'... the effect... on the economy was remarkably small just after a few months.”

0.30

Recession definitions and dating procedures are inherently fuzzy but recessions in practice separate clearly from normal times.

6 pointscentrality 1/5
  • The popular two-consecutive-quarters-of-falling-GDP definition of a recession is a not-bad generalization that fits some cases (December 2007 perfectly) but breaks down in others like 2001, especially since GDP is now measured monthly with zigzagging that creates dating challenges and revisions can flip a quarter from negative to positive.

    “the two quarters is a generalization that isn't bad... if you go back to the recession before that in 2001, that idea kind of breaks down... There was kind of a zigzagging of monthly real GDP... that still created a big challenge.”

  • The banking system as a whole has no choice whatsoever about its level of reserves—it is completely determined by the Federal Reserve—because if one bank invests its reserves, the funds are simply deposited elsewhere and the reserves land at another bank; it is a sealed system.

    “banks collectively have no choice whatsoever about their holding of reserves. It is completely determined on Constitution Avenue... If one bank decides to invest its reserves, the reserve just land at another bank. It's a sealed system”

  • The decades-long decline in real interest rates is a global phenomenon, not just American, driven substantially by slowing world economic growth including China; nominal rates were boosted in the 1970s by inflation and a double-whammy of monetary policy, peaked around 1981, then declined as inflation was controlled and real rates fell.

    “interest rates are global. This phenomenon all over the world. And world economic growth has slowed recently. It's not just the United States, but China, for example--a big factor is China. So, the decline in growth is one thing.”

  • The NBER business cycle dating committee mostly does nothing, only convening when members sense an expansion is ending, then often waiting a year or two for aggressive early data revisions to settle and to ensure a pause is not just a pause before resumed growth; in practice recessions have always been clearly recessions or clearly not, never marginal.

    “most of the time the committee does absolutely nothing... we wait to be sure that we're right. For one thing, as the revisions of the data tend to be quite aggressive over the first year. So, we wait for the data to calm down.”

  • The standard monetary policy of lowering interest rates was done about as aggressively as possible, while discretionary fiscal policy was not politically feasible, so the partial success of policy was not simply due to insufficient aggressiveness on the monetary side.

    “with respect to monetary policy, the standard policy of lowering interest rates was done about as aggressively as possible. Maybe not quite, but pretty close.”

  • What everyone calls a slow recovery should not be called recovery at all: the labor market followed a structurally normal recession/recovery path scaled up by a very large shock, but coincided with low productivity growth and weak capital formation since 2007, so the recovery is complete and all remaining disappointment comes from things unrelated to recovery.

    “The thing that makes everybody think that recovery was slow is not what we should be calling a recovery at all. It's what happens to productivity... The recovery is complete. But we are not where we'd like to be because productivity growth has been bad.”

0.24

Macroeconomic schools have converged; older theoretical divides like Keynesian-monetarist distinctions no longer structure serious academic disagreement.

2 pointscentrality 2/5
  • The notion of distinct schools of macroeconomics has largely disappeared, especially among the younger generation: 'Keynesian' means too many things, Lucas's rational expectations became fully integrated and is now considered old-fashioned, heterogeneous-belief models are sweeping the field with Chicago economists as enthusiastic as anyone, and it would be impossible to find a monetarist or old-fashioned Keynesian at a top university.

    “the idea of schools of macro has pretty much disappeared. Especially in the younger generation. It would be impossible find... an economist in a top university today who could be reasonably identified to be--well, especially a monetarist. But even a Keynesian in the old-fashioned sense.”

  • It is false that economists were clueless and had to start from scratch after 2008: the Bernanke, Gertler, and Gilchrist paper was already famous and formed the backbone of modeling financial effects on the economy, though the crisis did massively stimulate further integration of financial thinking into macroeconomic modeling faster than it would have happened otherwise.

    “it's just completely false to say that we were clueless and had never thought about this and that we had to start from scratch... the Bernanke, Gertler, and Gilchrist paper, which was already famous, became more famous as a result of being kind of the backbone of the modeling”