Sunday, December 6, 2015

The True Cause of the Business Cycle

Simon Wren-Lewis has a brief post on the business cycle where he points to policy (either fiscal or monetary) as the main driver of recoveries.  In his view, counter-cyclical policy is needed to correct for price rigidity, which is the market imperfection that gives rise to economic fluctuations.  I agree that policy is critical to the speed and shape of economic recoveries but do take issue with his view on sticky prices.  My work on time preferences reveals monetary policy and the bootstrapping problem affecting the economy as the true causes of the business cycle.

The reason for the business cycle is twofold.  First, banking and the monetary base may prevent income growth expectations by private agents, as measured by their time preferences, from properly aggregating into macro interest rates.  In a nutshell, as banks interface independently with borrowers and savers, and savers have the option to save by holding base money, there is no ex-ante requirement that desired borrowings should equal desired savings.  I define the aggregation error as the measure of such ex-ante disequilibrium.  The aggregation error causes interest rates to differ from the time preferences of private agents which results in either excess or shortage of desired borrowings.  This, in turn, causes the economy to either over or underperform compared to ex-ante income expectations and the price level to either rise or fall.  Such variance between actual and expected outcomes starts a feedback loop that powers the business cycle.

Desired savings and borrowings are at equilibrium only when ex-post, actual savings and borrowings have the same duration profile.  Duration measures the price sensitivity of a financial asset to changes in interest rates.  Assuming a financial asset with a fixed rate of interest, duration can be used as proxy for the asset life.  On the liabilities side, banks can transform duration by funding long-term financial assets with short-term liabilities.  The monetary base can also transforms duration as it can be used to purchase long-term assets.  On the asset side, the demand for such transformation comes from a particular group of private agents who want to hold their long-term savings in the form of money regardless of interest rates.  These are agents who have negative or zero time preferences as they expect declining or stagnant incomes.  In other words, even 0% interest either matches or exceeds the returns required by such agents per their negative or zero time preferences.  I refer to this component of money demand as asset money demand. 

Banks are not in position to meet asset money demand because their duration appetite is very pro-cyclical.  Banks go long on duration only in good times when few people expect declining incomes hence asset money demand is low.  In bad times, asset money demand rises, but the duration appetite declines as banks scramble for liquidity and capital.  Furthermore, under current regulatory regime, banks are required to duration-match their assets and liabilities in order to mitigate interest rate risk.  It is only the monetary base that can meet asset money demand; however, the monetary base is exogenous to the economy, and to the extent that it differs from asset money demand, the resulting positive or negative aggregation error induces economic fluctuations, changes in the price level as well as asset booms and busts. 

Under the Gold Standard, the monetary base was determined by the available supplies of gold hence there was an inherent disconnect between the monetary base and asset money demand.  This resulted in pronounced business cycles with significant swings in prices, wages as well as asset booms and busts.  Going back to Professor Wren-Lewis and mainstream macro, price rigidity cannot be the cause of the business cycle as there was no price rigidity to speak of under the Gold Standard.  Swings in prices and wages upwards of 20% were not uncommon.  It is precisely these price effects that marked the peaks and troughs of the cycle.  Peaks occurred when inflation expectations exceeded expectations of income growth and asset appreciation causing time preferences to decline.  Troughs occurred when deflation expectations exceeded expected declines in incomes and wealth causing time preferences to increase.

Under fiat money, central banks have full control over the size of the monetary base, which raises the prospect of zero aggregation error and no business cycle.  Zero aggregation error means that ex-ante desired savings equal desired borrowings.  Interest rates match income growth expectations by private agents as measured by their time preferences.  Prices remain unchanged, and the economy performs according to income expectations in the current period.  This is where the second cause of the business cycle comes into play.  Ex-ante income expectations are not necessarily consistent with full employment as the economy is subject to a bootstrapping problem - incomes have to be anticipated ex-ante in order to be generated ex-post (Chart 1).

Chart 1

People calibrate their spending plans based on their ex-ante income expectations and their desired borrowings and savings.  The money needed to accommodate such spending comes from either prior savings or new borrowings from banks (i.e. credit cards).  The spending plans by individual agents add up to aggregate demand which goes on to generate ex-post incomes.  Upon receipt of such incomes, people replenish their savings and pay-off their credit card balances (gross savings) such that only the desired borrowings persist into the next period.  Hence, desired borrowings always equal actual borrowings, which in turn equal the net increase in money supply and financial assets during the period or in other words, actual savings.  The take-away is that income expectations and net desired borrowings come first, while ex-post incomes and ex-post savings are simply residuals.  Furthermore, prices and wages are also residuals as the price level is determined by the aggregation error, and wages simply affect the allocation of actual incomes between wages, profits and rents. 

The important point here is that ex-ante income expectations do not have to be consistent with full employment.  Such expectations can be depressed for variety of reasons - demographics, unemployment, unsustainable debts, malinvestment or disinvestment, etc.  As a result, aggregate demand may not be sufficient for full employment.  Assuming endogenous process toward full-employment as mainstream macro does is equivalent to assuming that income expectations by the unemployed generate the aggregate demand necessary to get them employed.

The economy has historically relied on a positive aggregation error to overcome the bootstrapping problem.  The resulting excess borrowing boosts aggregated demand above ex-ante income expectations in the current period.  As the supply curve is fixed ex-ante by the available capital and technology, suppliers can respond to such excess demand by raising prices and employment thus moving the economy a step closer to full employment.  In other words, the economy bootstraps by relying on excess borrowing.  The problem is that borrowing is predicated on expectations of future income growth, but the Future is uncertain.  If such expectations are not realized, a positive aggregation error can quickly turn negative as agents borrow less and attempt to save more.  It is precisely such swings in the aggregation error that give rise to the business cycle.

Under inflation targeting, the economy continued to bootstrap by relying on a positive aggregation error.  Central banks correctly believe that by anchoring prices, they will eliminate economic fluctuations.  This is true because steady prices mean no aggregation error.  However, as I show above, steady prices may not lead to full employment.  In other words, interest rates consistent with price stability do not necessarily result in full employment.  Accordingly, due to their full-employment mandate, central banks defined price stability in terms of 2% inflation under the mistaken belief that inflation and unemployment are causally related.  Rather, it is the positive aggregation error needed to sustain 2% inflation that provides the boost to employment.  The problem is that the aggregation error works by boosting private debt and asset prices based on expectations of the Future that may or may not be realized hence the uncertainty and boom-and-bust nature of the economy has persisted.

From a business-cycle standpoint, the one difference between the Gold Standard and inflation targeting is that now central banks can prevent a negative aggregation error during recessions by expanding the monetary base.  This prevents prices from falling giving rise to the flawed notion of sticky prices.  More importantly, fall in prices is what triggered recoveries under the Gold Standard.  Under inflation targeting, central banks engineer recoveries by lowering interest rates below even depressed expectations such that private borrowing can power another recovery.  To use a simple metaphor, the prescribed medicine is what makes the patient sick in the first place.  With each cycle, the ever-rising burden of private debt depresses time preferences even lower pushing central banks toward the zero lower bound (ZLB).  At ZLB, continued expansion of the monetary base has little effect as agents with negative or zero time preferences are indifferent between holding money or government bonds.  In other words, at ZLB the central bank can prevent a negative aggregation error thus avoiding a depression but is incapable of inducing a positive aggregation error needed to engineer a recovery.  Accordingly, the economy performs according to depressed expectations with steady prices and high unemployment.

Can the promise of fiat money ever be fulfilled?  Is there a policy that can prevent the aggregation error and the business cycle that comes with it but also overcome the bootstrapping problem affecting the economy?  Absolutely! The first step is to understand the problem, which is what this framework attempts to do.  Empirically, this framework can explain nominal incomes, inflation and stock prices in the US going back to 1959 (based on data availability from FRED).  It provides a consistent account of the business cycle under both the Gold Standard and fiat money regimes including the stagflation of the 1970s and the period of inflation targeting.  Once you know the true cause of a problem, solutions will become self-evident.  I discuss one possible solution in my post on Money Demand Targeting and also in my post on how Money Demand Targeting can solve the Eurozone crisis.  This is where I share a firm belief with Professor Wren-Lewis that sensible policies can achieve steady economic growth at full employment with neither inflation nor asset booms and busts.  


  1. I think this labors within the false construct of saving funding lending. We know that's not how it works. Minsky explained the business cycle quite straightforwardly; it's a financial cycle, optimism begets pessimism, rising then falling prices of existing assets. Financial downturns cause real recessions. Wealth effect, with a significant theshold aspect. I think this series demonstrates that really clearly: Also makes sense intuitively: wealthy asset holders have much more discretion to change their spending levels suddenly.

  2. Thank you for the comment. I am a big fan of Minsky - actually, the framework I am proposing is an new micro-founded model that produces macro outcomes consistent with Minsky's financial cycle. However, unlike Minsky, who believed that instability was inherent to a free-market economy, I show that it is the monetary base that is the true source of booms and busts.

    If interest rates perfectly matched the ups and downs of income growth expectations (animal spirits), we would not have booms and busts. There are two necessary and sufficient requirements for such condition - banks have to duration-match their assets and liabilities and the central bank has to supply sufficient base money as to meet the demand for money by agents who expect declining incomes (agents with negative time preferences).

    Please know that I am not using the same micro-foundations as the New Keynesian model. The NK model uses ordinal utility (preference ordering rather than degree of preference). Ordinal Utility cannot be aggregated which constrains the NK model to implausible assumptions. Instead, I define utility in cardinal terms based on willingness to pay, which enables me to aggregate unlimited number of different agents and goods.

    I would love to correspond on these issues - please let me know if interested (my email in the about section)