Pages

Friday, September 25, 2015

Which is preferable - 4% Inflation Target or Negative Rates? I say Neither!

The UK has become a hot-bed of new ideas when it comes to macro-economic policy.  A recent speech by Andy Haldane, Chief Economist at BoE, puts forth two monetary policy alternatives for eliminating ZLB - either raising the inflation target to 4% or abolishing cash and charging negative rates on bank deposits.  The speech has already provoked a lot of discussion with commentators pointing to the disruptive effects of abolishing cash as well as alternative solutions such as heli drops and NGDP targeting.  My preferred ZLB solution is vesting the central bank with counter-cyclical fiscal tools in the form of temporary consumption and investment tax credits, which I discuss in my post on Corbynomics.  But what I would like to do here is comment on deeply-flawed 4% inflation target.
 
There are three reasons 4% inflation target is misguided.  First, inflation targeting is based on flawed micro-foundations; second, central banks cannot generate inflation at ZLB and last but not least, inflation targeting comes with the side effects of high private debt and asset booms and busts.  These risks are currently mitigated by low inflation target of around 2%.  However, by raising the target to 4%, central banks risk either more excessive debt and asset volatility or worst case, domestic capital flight and return of 1970s stagflation.



1. Flawed micro-foundations

Macro economist use the Euler Equation to model present and future consumption. For a great intro please refer to this post by Noah Smith, econ professor at Stony Brook University. In its simplest form, the maximization decision between two periods can be represented as follows: 
 
 

Per the Euler Equation, current consumption (Ct) is determined by the agent's expected future consumption (Cf), the current interest rate (r), expected inflation (i) and the agent's time preference (b), which is an exogenous input intended to represent the agent's preference for current over future consumption.

This simple formula captures a very intuitive understanding of the economy, which also underpins Andy Haldane's speech. High interest rates reduce current consumption while low rates give it a boost. At ZLB, central banks are constrained in their capacity to spur economic activity by lowering interest rates hence the idea of abolishing cash and charging negative interest rates on bank deposits. Alternatively, if central banks were to increase inflation expectations, consumers will be prompted to spend more today.  Intuitively, this all makes sense. The problem is that there is little empirical support for the Euler Equation (please refer to Noah Smith' post for more on the disconnect between the Euler Equation and observed consumption).  More fundamentally, such micro-founded model of consumption is subject to three flawed assumptions:  
  • Definition of Utility. Utility is not some unobservable measure of happiness and satisfaction. Rather, utility represents the maximum price an agent is willing to pay for a certain level of consumption. Therefore, utility is not an ever increasing function at an ever declining rate as assumed by the Euler Equation (logarithmic utility assumed in the above example). Instead, the utility function can be approximated as illustrated in Chart 1 below where C stands for consumption, Qmin for the minimum real consumption necessary to provide for the most basic needs and P for the price level.   At low levels of consumption, utility approaches infinity as agents are willing to pay any price for life-saving necessities.  At Qmin all necessities are satisfied hence the utility function starts to increase as agents switch to luxury goods.  At high levels of consumption, utility approaches the spending constraint as there are fewer lower-income marginal buyers that can bid prices below what a millionaire would be willing to pay.  The spending constraint represents the fact that the maximum price an agent is willing to pay can never fall below the price actually paid. 
Chart 1
  • Maximization Behavior. Agents do not seek to maximize the absolute level of utility as assumed by the Euler Equation but rather their utility gains. Utility gains represent the difference between the maximum price an agent is willing to pay for a certain level of consumption and the price actually paid. Under this interpretation, the actual price paid represents the opportunity cost of all alternative consumption baskets that the agent chose not to consume. In other words, just like companies seek to maximize profits, so do consumers seek to maximize utility gains. 
        
  • Time Preference. Time preference is not an exogenous, always-positive input into the Euler Equation that discounts future consumption into the present. Instead, time preference measures the change in utility gains due to delayed consumption. Utility gains can be approximated as the inverse of consumption.  As the level of consumption is determined by period incomes, time preference is endogenously determined by income growth expectations. If agents expect declining incomes, they will expect higher utility gains in the future hence they will value future consumption more.  If agents expect rising incomes, they will expect declining utility gains and will value future consumption less. Accordingly, agents make saving and borrowing decisions as to smooth-out the utility gains of their current and future consumption.
This model of time preferences represents a radically different approach to understanding consumption, borrowing and saving decisions.  It is the basis for the aggregation error framework for modeling the macro economy, which I discuss at length in my post on Time Preferences, Interest Rates and Money Demand Targeting.  For starters, income growth expectations as measured by the time preferences of private agents have to be brought into the intertemporal maximization decision. For  a moment, let's assume no inflation. 


 
Time preference (T) measures the expected growth in incomes (Y). If an agent's time preference today (Tt) exceeds the observed interest rate (rt), she will be a borrower (B>0).  Conversely, if the interest rate exceeds the agent's time preference, the agent will be a saver (S>0). If the agent expects her income to decline, her time preferences will be negative. Such agent values consumption tomorrow more than consumption today (the proverbial “saving money for a rainy day”).  Since interest rates cannot fall below 0%, such agent will always be a saver.

Now let's consider the impact of inflation expectations.  While inflation does erode the future purchasing power of savings, it also erodes the purchasing power of incomes.  Accordingly, inflation expectations reduce both time preferences and interest rates.  In other words, you cannot stimulate consumption by raising inflation expectations unless you increase income growth expectations by more than the increase in expected inflation.  To summarize: price expectations on their own are meaningless.  Instead, to project their impact on the economy, price expectations should always be netted out of income growth expectations.


Those who advocate for higher inflation target as cushion against ZLB, should be careful what they wish for. As I demonstrate above, inflation expectations cannot be considered in isolation of incomes. At best, 4% inflation target will have no effect as both expected real interest rates and time preferences are reduced by the same amount.  However, the more likely scenario is that higher inflation target will not be matched by higher income expectations resulting in depressed time preferences.  The combination of low income expectation with high inflation may prompt domestic capital flight into gold, commodities and FX as agents attempt to preserve wealth.  Under such conditions, it is not unreasonable to expect a return of 1970s stagflation.

2. Asset Money Demand - the key to both disinflation at ZLB and 1970s stagflation.
 
Agents with negative time preferences play a very critical role in the economy.  These are people who expect declining incomes.  Such agents are willing to save regardless of interest rates, but even more importantly, they are willing to hold their long-term savings in the form of money.  It is simple arbitrage - even 0% return on cash exceeds their required returns per their negative time preferences.  I refer to this component of money demand as asset money demand. It stands for the store-of-wealth motive and represents long-term savings held in the form of money by agents with negative or zero time preferences.

It is important to note that banks are not in position to satisfy asset money demand with bank money.  Only to the extent banks go long on duration, can bank money satisfy asset money demand.  However, the supply of borrower liabilities, which is the raw material for money creation by banks, is inversely related to the time preferences of private agents.  In other words, when time preferences fall, asset money demand rises as more agents attempt to hold their long-term savings in the form of money but the supply of bank money declines as fewer agents are willing to borrow.  Furthermore, banks appetite for duration is also very pro-cyclical and last but not least, regulators require banks to duration-match their asset and liabilities, which further constrains banks' capacity to meet asset money demand by going long on duration.

It is only through the exogenously-supplied monetary base that asset money demand can be fully satisfied.  The aggregation error measures the difference between the monetary base and asset money demand which, in turn, equals the difference between ex-ante desired borrowings and savings.  Positive aggregation error, meaning that the supply of base money exceeds asset money demand, pushes interest rates below time preferences.  This results in excess desired borrowings causing the economy to perform above expectations and prices to increase.  A negative aggregation error, meaning that the demand for asset money exceeds the supply of base money, pushes interest rates above time preferences.  This leads to a shortfall in desired borrowings causing the economy to perform below expectations and prices to decline.  It is precisely this variance between actual and expected outcomes in the current period that gives rise to the business cycle, inflation and asset booms and busts.  

According to this framework, changes in the price level are residuals of the ex-ante disequilibrium between money supply and demand as measured by the aggregation error[1].  For a point of reference, Chart 2 compares actual inflation in the US with projected inflation per the aggregation error framework. 

Chart 2

Now, let me go back to the question of whether central banks can eliminate ZLB by raising inflation expectations.  At ZLB, the aggregated time preferences of private agents fall into the negative.  Central banks can prevent a shortage of base money and the corresponding price declines by rapidly expanding the monetary base to meet rising asset money demand.  However, continued expansion of the monetary base via purchases of government securities cannot generate excess supply of base money as asset money demand continues to rise as well.  Simply, agents with negative or zero time preferences are indifferent between holding money or risk-free financial assets.  In the absence of aggregation error, inflation remains subdued and the economy performs according to depressed expectations.  In other words, central banks are unable to generate positive aggregation error at ZLB, hence the lack of inflation.  Without actual inflation, the task of raising inflation expectations becomes doubly difficult.  If anything, inflation expectations de-anchor from target and may begin to decline (which, of course, is a positive in the face of stagnant income expectations as I discuss in point 1 above).

Here I need to make an important caveat.  The analysis above assumes that domestic agents perceive the domestic currency as a monetary reserve capable of satisfying their asset money demand, which is not necessarily the case.  In the 1970s, agents in the developed world continued to seek the safety of gold to satisfy asset money demand.  To this day, emerging and developing markets suffer domestic and foreign capital flights during recessions, which subjects such economies to the double scourge of high inflation and high unemployment.  Even though such countries issue their own currency, they lack monetary sovereignty because domestic agents choose foreign over domestic currency to satisfy their asset money demand.  This should act as a word of caution to advocates of abolishing cash and charging negative rates on deposits.  Asset money demand is real, and people with negative time preferences will seek out other alternatives if policy makers deliberately undermine the reserve status of the domestic currency.  

3. Inflation targeting, private debt and asset booms and busts.

Inflation-targeting central banks subscribe to the view that 2% inflation target is good for the economy because it stimulates economic activity, keeps unemployment at bay and provides cushion against ZLB.  In point 1 above, I address the flawed micro-foundations of this argument.  In point 2, I discuss the incapacity of central banks to generate inflation at ZLB, which means that this supposed cushion is nothing more but a sugar pill.  Even more damning evidence against the inflation-targeting mindset is the complete breakdown of the Phillips Curve since the Great Recession, which I discuss at length in my post on money velocity and whether the Fed should raise interest rates.

The question remains - what has been the transmission mechanism over 25 years of inflation targeting? According to the aggregation error framework, the source of inflation is ex-ante disequilibrium between money supply and demand as measured by the aggregation error.  In the absence of aggregation error, Say's law holds, and an economy performs according to expectations with no inflation.  It is an entirely different question whether such expectations are consistent with full employment.  Expectations can be depressed for variety of reasons such as war or bad weather, ageing demographics, unsustainable debt, asset busts, etc.  Under such conditions an economy can experience consistent unemployment [2].

Inflation-targeting central banks have had to consistently maintain a positive aggregation error in order to achieve the 2% inflation target.  Suppressing interest rates below time preferences means that asset prices rise[3] and private agents take on more borrowing than they otherwise would per their income expectations.  In the medium term, inflated asset prices and excess private borrowing depress time preferences as asset bubbles burst and private debt becomes unsustainable.  The tried-and-true response by central banks has been to lower interest rates even more, which initiates a new cycle of private borrowing and inflated asset prices.  This vicious circle of debt and asset bubbles inexorably sets inflation-targeting central banks on the path toward ZLB.

In conclusion, raising the inflation target to 4% is misguided.  Such proposal is based on flawed understanding of inflation and how it impacts agent behavior.  The best case scenario would involve even more private debt and even more asset volatility.  Under the worst case scenario, domestic currency will lose its reserve status giving rise to 1970s stagflation and depriving policy makers of the monetary sovereignty needed to conduct counter-cyclical fiscal and monetary policy.


    

 
[1]  The aggregation error framework can be viewed as re-interpretation of the Quantity of Money Equation whereby velocity is a measure of ex-ante money demand and the causality in the ex-post QP=MV identity flows from MV with prices and employment being the residuals.  In other words, since capital and technology are fixed ex-ante, businesses can respond to ex-ante change in aggregate demand caused by the aggregation error only by changing prices and employment in the current period.

[2] There is no endogenous process driving the economy toward full employment.  According to the aggregation error framework, ex-ante income expectations are the main driver of economic performance.  Assuming an endogenous process toward full employment is equivalent to assuming that ex-ante income expectations by the unemployed can generate the aggregate demand necessary to get them employed. 

[3] Expected returns on financial assets are an important component of time preferences. However, market prices of financial assets rely on interest rates to discount such expected future returns. As a result, suppressing interest rates below time preferences inflates the price of financial assets.


No comments:

Post a Comment