Sunday, June 29, 2014

Thoughts on Inflation/NGDP Targeting. The case for Money Demand Targeting.

There have been repeated calls for central banks to increase inflation targets (most recently by Paul Krugman here and here) or even switch to Nominal GDP "NGDP" Targeting (most recently by Wolfgang M√ľnchau in the FT; also look up Scott Sumner, Lars Christensen).  The thought is that such policies will prevent "lost decade(s)" of secular stagnation at the zero lower bound and help the economy make up the output gap caused by the Great Recession.

I first approached the subject in a jovial manner when I recounted a recent conversation with my wife (using broad poetic license).  On a more serious note, this is an attempt to list the reasons why Inflation and NGDP Targeting may be misguided.  Also, I will describe a new paradigm for monetary policy - Money Demand Targeting, which can provide for sustained growth without distortions such as inflation and asset booms and busts.

1. Under a credible central bank, inflation is subject to a time lag. As a result, Inflation Targeting has been the primary driver behind asset bubbles over the last 30 years.

I've observed repeatedly that after Chairman Paul Volker established the Fed's credibility in the early 1980's, inflation no longer reflected real-time changes in the value of money. Instead, over the last 30 years changes in asset prices have been the more accurate measure with asset booms and busts closely matching imbalances between the supply and demand for money.  I demonstrate this point below with the chart from my post on Endogenous Money ISLM (Ma in the chart stands for asset money demand).
Chart 1 (Source: FRED)

There could be a number of explanations.  First, a credible commitment to price stability by a central bank could act as an anchor to inflation expectations.  Such anchor could also discourage arbitrage and stabilize exchange rates and consumer behavior creating self-fulfilling expectations.  Under such conditions, excess money supplies would be channeled toward asset markets driving asset booms, which initiates a feedback loop of asset appreciation and improved confidence as I describe in my post on rational agents and irrational macro outcomes.  The chart below clearly demonstrates the anchoring effect with expectation never straying far from the 2-2.5% anchor and actual inflation converging over time with expectations:

Chart 2 (Source FRED)

Another important consideration is the unequal distribution of money balances. Chart 3 below estimates money distribution among US households.  It shows that households in top income percentile hold between 40% and 50% of all money balances held by US households.

 Chart 3 (Source: 2010 Survey of Consumer Finances by the Federal Reserve)

Central banks achieve their interest rate targets by manipulating money balances held by the public.  However, unequal distribution of money balances places a constraint to monetary policy.  Wealthier households have a lower propensity to consume, and they are more likely to manage cash preferences by changing asset allocations rather than changing spending decisions.  Accordingly, an excess supply of money will affect asset prices first as money balances begin to churn among high income households giving rise to an asset bubble.  Only when higher asset prices begin to attract more investment and the associated wealth effect gives a boost to confidence in the lower income percentiles, will the effects of monetary policy begin to build up inflation pressures in the real economy.  In addition, asset appreciation pushes time preferences above interest rates, which gives a boost to borrowing.  People borrow more than they otherwise would have because the cost of money is below time preference.  Such debt overhang prolongs the downward leg of the cycle and makes it more painful.

Due to unequal distribution of money balances, inflation signals are subject to a time lag.  As a result, Inflation Targeting is the true cause for asset booms and busts.  First, central banks inflate the monetary base above money demand for an extended period as to meet the inflation target.  This fuels asset bubbles and the demand for borrowing.  Later as inflation catches up and overshoots the target, central banks start to put on the breaks deflating the bubble and initiating the downward leg of the cycle.  To sum-up: due to delayed signaling, Inflation Targeting has been the main cause for the two asset boom-and-bust cycles of the 1990's and 2000's, a period ironically called "The Great Moderation".

2. Inflation comes with significant welfare costs. Inflation Targeting could be a contributing factor to the rise in wealth and income inequality.

Inflation comes with high social costs because it does not affect people equally.  The idea that nominal wages magically adjust upward to absorb the impact of higher prices is simply wrong.  In fact, due to high job search costs, globalization and the declining influence of unions, workers are at a disadvantage while employers enjoy pricing power.  This creates upward wage rigidities, which could significantly delay wage adjustments for inflation.  When you add a minimum wage regime in the United States that is not indexed to inflation, it becomes quite evident that the brunt of inflation costs are being inflicted on low and middle income wage-earners.

A recent paper by the St. Louis Fed identifies another social impact.  Lower and middle income households keep a greater percentage of their wealth in cash/bank deposits so they suffer proportionally higher loss of wealth due to inflation.  Even more importantly, a moderate increase in inflation causes such households to actually cut spending and increase holdings of cash due to their use of money as self-insurance against consumption shocks.  The paper estimates that a moderate rise in inflation from 0% to 10% can actually lead to consumption cuts in the range of 2-3%.

The counter argument is that welfare costs due to inflation are offset by welfare benefits due to low unemployment.  The suggested trade-off between inflation and unemployment is described by the Phillips Curve.  However, the Phillips Curve is not without its challengers (Roger Farmer, economics professor at UCLA, discusses the history of the Phillips Curve and its flaws).  Fundamentally, Endogenous Money ISLM reveals the workings behind the Phillips Curve.  The curve describes coincidence rather than causality.  Both inflation and unemployment are dependent on asset money demand, which is the true driver of macro-imbalances.  Under a credible central bank or the gold standard, unemployment is positively related to asset money demand while inflation is negatively related, which gives rise to the Phillips Curve.  Under fiat money regime without a credible central bank, both inflation and unemployment are negatively related to asset money demand hence the paradox of stagflation and the failure of the Phillips Curve to describe reality.

The good news is that there is another way.  If policy makers develop tools to measure and manage asset money demand, we could achieve low unemployment without having to rely on inflation.  There is a fallacy in the premise that only low levels of inflation indicate an economy operating at potential with no output gap.  Rather, this is the current state-of-the-art when it comes to monetary policy.  Money Demand Targeting, as I discuss below, is that new horizon, a new frontier which promises full employment without the welfare costs of inflation.

3. At the lower zero bound, central banks are unable to induce inflation.

Probably the most obvious challenge to Inflation Targeting is the fact that central banks cannot induce inflation at the lower zero bound as demonstrated by the last 5 years of extraordinary efforts to pump money into the economy with little to no inflation.  The experience in the US is probably most telling since the Fed pulled all the stops and engaged in not one, not two but three rounds of QE that added trillions to its balance sheet.

Again, Endogenous Money ISLM provides the answer.  The lower zero bound is caused by negative expected returns, which people arbitrage by holding cash at 0%.  Accordingly, demand for money becomes infinite (a concept first proposed by Keynes as infinite liquidity preference).  To rephrase in simple terms: people expect to be poorer in the future so they stash money for a rainy day.  I re-post Chart 4 below, which is quite telling - asset money demand rises drastically as time preferences approach zero.

Chart 4 (Source: Endogenous Money ISLM)

That is not to say that QE has been without benefit, quite the opposite.  At the onset of the crisis, negative expected returns prompted households and corporations to dramatically increase holdings of money.  By standing ready to supply new money balances, the Fed has prevented drastic cuts to spending and deflation.  Basically, active monetary policy has averted the depressionary downside of the crisis; however, subsequent efforts to induce nominal and real growth have been offset by continued increases in asset money demand.  As a result, we've seen three rounds of QE with the Fed pumping ever larger amounts of money in the hope that inflating asset values will pull the economy out of the slump.  However, real growth will occur only when time preferences lift-off above zero and demand for asset money declines.

This is a good place to say a few words about fiscal policy and deficit spending in particular.  The fundamental point is that monetary policy and government deficit spending are substitutes because both policies affect the supply of money.  The question is whether you want to meet the massive demand for asset money with new reserves printed by the Fed or deficit spending by the government.  Where the policies do differ are their distributional effects and the proportion of government to private spending, but that is a subject for a different post.  The fact of the matter is that at the zero lower bound both policies are equally constrained by the massive increase in money demand.  To overcome such constraint, you need something on a super-massive scale as in QE I, II and III or World War II.

4. Thoughts on NGDP Targeting.

Proponents of NGDP Targeting contend that monetary policy has not been accommodative enough.  Developed economies experienced a drop in nominal GDP at the beginning of the crisis and are yet to reach potential level assuming some steady nominal growth path.  Chart 5 below illustrated the US output gap between current GDP and its potential level as estimated by the Congressional Budget Office.

Chart 5 (Source FRED)

According to NGDP Targeting, central banks in setting monetary policy should take cues from the output gap between actual GDP and potential GDP based on some predetermined nominal growth path.  Inflation Targeting on the other hand lets bygones be bygones - if the economy experiences below-target inflation but then reverts back to target, the central bank has fulfilled its mandate without having to concern itself with making up for past output gaps.  If, instead, market expectations were to be anchored to nominal GDP commitment by central banks meaning that easy money will persist until the output gap is eliminated, such nominal drops would not have occurred in the first place.

The problem is that central banks cannot induce inflation and nominal GDP growth at the lower zero bound.  As we saw above (Point 3), demand for money becomes infinite.  Regardless of how much money is pumped into the economy, such newly-minted reserves are quickly stashed away in mattresses and bank deposits as evidenced by the dramatic rise in excess reserves since the recession.  Now, there is a good argument that the Fed may have been slow to act prior to the Lehman collapse and could have done more to avert it.  However, the fact remains that after Lehman filed for bankruptcy, which was the shock that pushed us to the lower zero bound, the Fed pulled all the stops pumping trillions of new reserves, bailing out financial institutions and even putting government backstops to private financial markets.  Could the Fed have done more?  I don't know.  It is simply too hard to prove such counter-factual.

If you accept the premise that output gaps will occur, which is bolstered by the fact that asset markets are subject to booms and busts, then the practical implication of NGDP Targeting is that central banks should be prepared to accept much higher levels of inflation in the short-term.  By definition, you cannot make up for lost time.  The potential output lost during the long years of high unemployment and idle factories is gone forever.  The only way to get back to potential GDP is by raising the price level.  The benefits of making up such nominal output gap through inflationary expansion are debatable, and as illustrated above, there are significant risks and costs.  To sum-up: monetary policy is unable to induce inflation and nominal GDP growth at the lower zero bound when it is most needed.  Later on, when monetary policy is in a position to induce inflation, such higher nominal GDP growth is no longer needed.

5. New Paradigm - Money Demand Targeting

I hope by now it is starting to become clear that we need a new paradigm of money, one that focuses on asset money demand.  I explain the macro significance of money demand in the following post: Endogenous Money ISLM.  Also, you can find micro-foundations and detail workings of Money Demand Targeting here: Efficient Markets, Rational Agents and Asset Bubbles.

Money Demand Targeting has three building blocks.  First it strives to understand money velocity.  Money velocity has much deeper meaning than what is generally deduced from the quantity of money equation.  Velocity reveals asset money demand using a modified quantity of money equation (Mt stands for transaction money demand, Ma for asset money demand and Vt is the velocity of transaction money). Transaction money velocity Vt is exogenous because it is determined by things such as payment technology and purchase friction.  By assessing Vt, policy makers can derive asset money demand Ma

The second building block is understanding the sources of money demand.  The source of transaction money demand, is the endogenous money created by banks - money supply (M) less the monetary base (MB).  The source of asset money demand is the time preference in the economy.  Time preferences (T) and interest rates (r) are two distinct measures of expected future returns.  Time preferences measure those expectations in the real economy as expressed by actual spending and investment decisions.  Interest rates measure those same expectations as expressed in terms of the supply and demand for money.  A note to mainstream economic thought: Keynes would have referred to time preference as the opposite of uncertainty and Wicksell would have called it the Natural Rate of Interest.

The fundamental insight from Endogenous Money ISLM is that time preferences (T) in the real economy and interest rates (r) in financial markets match only when the monetary base (MB) equals asset money demand (Ma).  This framework is consistent with endogenous money creation by banks.  Furthermore, it reveals the feedback loop which is at the heart of the business cycle:

If return expectations increase due to technological innovation, time preference will increase as well which reduces the demand for asset money.  Unless the monetary base declines to match the decline in asset money demand, time preferences will rise above interest rates causing assets to appreciate and consumption to increase.  Banks can meet the excess demand for borrowing by simply expanding their liabilities and creating new money in the process.  This puts in motion the self-reinforcing cycle of asset appreciation and increased economic activity. The upward leg of the cycle ends when the combined effect of diminishing returns associated with higher levels of investment and increased inflation expectations due to higher consumption overwhelm positive return expectations setting the stage for the downward leg of the cycle.  Declining time preferences lead to rapid increase in asset money demand.  Unless the monetary base expands, this triggers asset liquidation and rapid decline in economic activity as agents cut spending and investment to meet their demand for asset money.  Banks can accommodate the extra supply of savers by expanding reserve holdings at the central bank provided that the central bank is willing to oblige by expanding its own balance sheet.  Under a gold or fixed reserve standard, such condition leads to deflationary contraction.

The third leg of Money Demand Targeting is a set of new tools that can impact time preferences as opposed to interest rates and in the process enable central banks to manage the demand for money.  Another way to look at time preferences and interest rates is in terms of the utility of money (basically, answering the question how much utility can money buy).  Utility of money is always relative to wealth.  If you expect to be richer in the future (positive time preference), spending $100 dollars today will bring a lot more utility than spending the money in that blissful future when all of your needs will be satisfied.  Hence, if you are a saver, you will demand interest to compensate for the loss of utility associated with delayed consumption.  If you are a borrower, you will not be opposed to paying interest due to your utility gain associated with spending money today.  On the flip side, if you expect to be poorer in the future (negative time preference), spending money in the austere future will bring a lot more utility than spending money today when you are relatively well-off.  As a result delayed consumption leads to utility gain by savers hence they are not opposed to holding cash at 0% (saving money for a rainy day).  By the same token spending borrowed money today leads to utility loss causing borrowing to plummet even if rates fall to 0%.
  • Consumption/Investment Tax Credits
    Consumption taxes are a powerful tool for managing the comparative utility of money over different points in time.  Suppose central banks had the capacity to fund temporary consumption and investment tax credits.  If negative time preferences drive interest rates to zero, central banks can initiate massive stimulus by announcing a 10% consumption tax credit.  In one stroke, the utility of current spending rises in comparison to future spending, which basically does away with negative time preferences at the lower zero bound.  The Fed pumped trillions of dollars through QE with little effect on nominal growth and inflation.  Money stimulus delivered through a consumption tax credit would be much more effective because it goes straight into the real economy as opposed to QE money that can be stashed away by the rich or funneled into asset bubbles.  More importantly, the fact that time preferences are back in positive territory starts to put downward pressure on asset money demand and break open the liquidity trap.

    Important Note: armed with consumption and investment credits, policy makers face no limit in their capacity to manage the utility of money and manipulate time preferences in the process.  Just as important: central banks will reap a political benefit - they will be seen as helping Main Street as opposed to Wall Street.

    Such credits could also be very effective in managing foreign money flows that often place constraint on domestic monetary policy especially in developing economies.  When foreign money is easy and abundant, higher consumption taxes can dampen consumption-fueled trade deficits while investment credits can direct such financial flows into domestic capital accumulation.  When foreign money dries out, higher consumption credits can soften the impact on the domestic economy. First and foremost, central banks in developing markets have to establish credibility.  Accordingly, they have very little flexibility in deviating from their respective monetary target (either steady exchange rate or low inflation).  Creating a new tool set based on consumption/investment credits will greatly enhance their ability to achieve domestic macro-economic goals while staying true to their monetary targets.

    Currency unions such as the Euro can reap a further benefit.  Unlike the United States where state borders do not constrain the movement of labor and capital, Europe is still a continent defined by national borders, language barriers and ethnic divisions.  One-fits-all monetary policy by the ECB cannot accommodate the divergent needs of the periphery and the core.  National consumption tax credits funded by the ECB can customize monetary policy to the needs of each member country.  More on that here.
The objective of Money Demand Targeting is sustained growth without distortions such as inflation and asset booms and busts.  Rather than relying on low levels of inflation as indicator of economy at full potential, money demand targeting seeks to achieve a balance between the supply and demand for base money.  Such condition is characterized by an equilibrium between saving and borrowing where for every non-bank saver there is a non-bank borrower.  This eliminates the possibility of both output gaps and excess demand and removes the lower zero bound constraint.  Under this regime, central banks will not fuel asset bubbles by manipulating interest rates.  Instead, interest rates will be a true representation of expected real productive growth in the economy.  Asset prices will reflect relative performance rather than fluctuations in the value of money as is the case today.  Finally, this is a world without inflation where money is truly a neutral medium of exchange as opposed to the powerful cyclical force it is today.

* Note to reader: all research is based on US monetary statistics. Due to its dominant global reserve status, the US dollar behaves very much like gold under the gold standard.  This framework may not be directly transferable to other currencies because agents in other countries have a choice of FX reserves when attempting to meet their asset money demand.  I am in the process of developing Money Demand framework for an open economy with a choice of FX reserves.

No comments:

Post a Comment