Sunday, May 5, 2013

The Seen and Unseen: Structural Budget Deficits Edition

On Friday I discussed the cyclically-adjusted U.S. budget deficit and asked any Keynesian to reconcile its decline with the steady growth of aggregate demand. Robert Waldman graciously replied, but he really didn't answer my question. His response focused on the pace of the recovery and changes in the overall budget balance. My question was about the structural budget balance. This distinction is an important one. So let me try this one more time.

The structural budget balance is the best way to gauge the stance of fiscal policy, as noted by Paul Krugman:
[M]easuring austerity is tricky. You can’t just use budget surpluses or deficits, because these are affected by the state of the economy. You can — and I often have — use “cyclically adjusted” budget balances, which are supposed to take account of this effect. This is better; however, these numbers depend on estimates of potential output, which themselves seem to be affected by business cycle developments. So the best measure, arguably, would look directly at policy changes. And it turns out that the IMF Fiscal Monitor provides us with those estimates, as a share of potential GDP...
Here is the IMF's cyclically-adjusted or structural budget balance for the United States:

So what does this figure tell us? It shows that fiscal policy, independent of business cycle influences, has been tightening since 2010. It has gone from a deficit of 8.5% in 2010 to an expected one of about 4.6% in 2013.  In other words, the reduction in the federal budget deficit over the past three years is more than just the government adjusting its balance sheet in response to improvements in the private sector's balance sheet. It is also the result of explicit policy choices to impose fiscal austerity. And these explicit policy changes have been relatively sharp.

Many observers have overlooked the implications of this structural austerity experiment. Three years of explicit fiscal austerity in a depressed economy should, all else equal, lead to even more economic weakness. But it has not. Nominal GDP growth--a proxy for aggregate demand (AD) growth--has been remarkably steady. There is no evidence AD over the past three years has been adversely affected by this austerity. Friday's job report underscores this point.

So what explains this development? The answer is not that fiscal policy has no effect, but rather that all else is not being held equal for U.S. aggregate demand growth. Specifically, Fed policy has effectively responded to the fiscal austerity, Eurozone shocks, China slowdown shocks, and other shocks to AD. Though this is a great accomplishment, it is far from adequate and is ultimately frustrating to watch. For it speaks to both the power and shortcomings of current Fed policy.

It is not surprising to me that Keynesians and other observers fail to see this structural austerity. The Fed has offset it over the past three years and therefore kept it out of sight, out of mind. The ECB, on the other hand, has not and so it is more apparent to observers. But just because it is not seen, does not mean it is not there.


  1. Let's get truly Keynesian...
    Aggregate demand is a function of aggregate supply. They both meet at an "equilibrium point" in the AS-AD model. So if aggregate supply increases, it is calculated and assumed that aggregate demand is increasing. There is an error to this thinking.

    I use a different measure of aggregate demand called Effective demand, which is a primary concept of Keynes' work. Look at the graphs at this link...

    You will see that Effective demand is actually falling, not increasing. It is to be expected that effective demand will decrease through the recovery. According to Keynes, eventually the effective demand curve will cross the aggregate supply curve. At which point, effective demand will limit aggregate output. We are heading toward that point.

    There is an important difference to realize... In one hand aggregate demand always increases with aggregate supply in the AS-AD model, but in the other hand effective demand can be decreasing in relation to aggregate supply. My model shows thus better represents Keynes' fundamental concept of effective demand.

    In the second graph of the above link, you will see that in the past year, aggregate supply has been shifting right and the effective demand curve has been holding steady. But as aggregate supply shifts right, effective demand is decreasing due to its slope.
    The economy will use up spare capacity for labor and capital up until the effective demand curve crosses aggregate output. (Keynes' concept).

    The economy is naturally expanding within its limits, but getting closer to those limits.

    So your question assumes a growth in aggregate demand because aggregate supply is growing. That assumption is fundamentally an error. It only appears to you that aggregate demand is increasing because aggregate supply is increasing. However, the true issue is not that aggregate demand is increasing, but rather that effective demand is naturally decreasing as aggregate supply increases in the recovery.

    Eventually aggregate output will be limited by effective demand. At which point, someone would say that aggregate demand is finally limiting aggregate supply. But we can see now that it is a natural process of a decreasing effective demand. We can also estimate the point at which effective demand will limit aggregate output... somewhere between $14.000 and $14.200 trillion real GDP.

    It may be hard for Keynesians to answer your question because they use the same AS-AD model, and thus share your fundamental error in understanding the difference between aggregate demand and effective demand.

    You will only find my effective demand-aggregate supply model on my blog. It has not been officially published yet. But my model explains the error of your question.

    1. David, I watched a video with you and Scott Sumner last night. I thought about many things you said in relation to inflation volatility, inflation self-correction, demand liquidity, returning to and staying on long term trends... I agree with you on many of your insights, but I have gone in a different direction to solve these issues and I think I am right.

      The real issue is weak demand, but that will only be solved by higher labor share of income. Monetary policy is ineffective for that. We lack a true mechanism to move liquidity from capital to labor. That is where your model will fail.

      I have four posts to address your video...
      1. I have updated the AS-ED model. Nominal GDP targeting would be ineffective against declining effective demand, unless you could somehow raise unit labor costs. I don't see how you can insure that. Inflation is unable to push up against a declining effective demand when capacity utilization is still low and unemployment high. The dynamics you need for nominal GDP targeting are not there.
      2. To me we have returned to the long-run natural trend of the economy after years of false bubbles. Effective demand in the link's graph was artificially inflated for many years. Effective demand has come back down to normal and will help keep real GDP at the historic trend line. Whereas you seem to say that real GDP should head back up toward what I would call a "false potential real GDP". If you try to push nominal GDP back up, how will you increase labor's liquidity without reverting to easy credit and bubbles?
      3. My model of monetary policy shows that labor share of income is critical to a viable Fed funds rate. How would nominal GDP targeting address the issue of a low labor share of national income? Through expectations of higher future income based on inflation? Wouldn't it be better to have higher future income based on higher real wages? How could you keep real wages from falling more?
      4. I agree with you about allowing inflation to be volatile. It is healthy for the economy. It re-balances unit labor costs which then raise effective demand. My model does not calculate a difference between the inflation rate and the inflation target and then try to automatically close that gap like the Taylor rule. I think that the Taylor rule is ultimately bad for the economy. My equation embeds a relationship between unit labor costs and inflation through the labor share variable. If unit labor costs rise relative to inflation, then labor share increases leading to an increase in the Fed rate. On the other hand, if inflation is rising, but unit labor costs are falling in relation, then labor share is falling. My model would reduce the Fed rate in the face of rising inflation to preserve effective demand. To do as the Taylor rule does, is to damage the re-balancing of unit labor costs through the business cycle, which weakens effective demand on average through the cycle.

      These are some of my thoughts since watching your video... but there is more going on in your presentation too.

  2. David, I think you are re-opening the famed FM/AM debate of the 1960s. This time around is as unlikely to achieve resolution as the first time.

    What next ? Stones vs The Beatles ?

    1. Note sure I follow you. Are you saying the FM/AM 1960s debate is akin to the monetary policy vs. fiscal policy discussion above?

  3. David, I see that Scott is talking about targeting labor share for monetary policy now. That is what my model is about. According to the equations and the graph, if labor share of income drops below a certain point, monetary policy is ineffective. If you can raise it back above that point, monetary policy will have traction again. I have a model ready for you to test with data.

  4. How exactly has the Fed offset fiscal austerity in the U.S.? If the Fed really were boosting aggregate demand, and hence nominal GDP, I would expect that the NGDP curve would change slope around shifts toward accommodative monetary policy (e.g., September 2010 with QE2, November 2011 with Operation Twist, September 2012 with QE3). But the slope of the NGDP curve is (almost) flat ever since the start of the "recovery" in 2009.