Instrument Rules vs Targeting Rules

Instrument Rules vs Targeting Rules Instrument rules vs targeting rules? Should Central Banks commit to a simple instrument rule such as the Taylor Rule? How well does such a rule explain Central Bank behaviour? Do targeting rules provide a more compelling alternative? The Taylor rule has had a wide-reaching effect on the literature surrounding monetary policy design. It is a simple instrument rule which aims to show how interest rates should respond to two economic indicators: Inflation and Output. This simple rule has led to the â€Å"Taylor Principle† which has been said to be useful in guiding policy design. Many studies have been done to determine whether or not the Taylor rule appears to guide monetary policy in many central banks. However there has been a vocal group that criticises the Taylor Rule and instrument rules in general as being inflexible and not allowing for discretionary measures, this has sparked lively debate between the two sides of the debate. This essay will examine the literature surrounding the Taylor Rule and analyse the good and bad aspects of the rule. In addition to this the empirical studies examining the Taylor Rule will be discussed. Furthermore a brief look targeting rules will provide a useful counterpoint to the analysis of instrument rules. Literature surrounding the Taylor Rule Before discussion of the literature surrounding the Taylor Rule can begin it is necessary to define what Central banks aim to achieve with monetary policy. Most agree that inflation targeting is a key concern for central banks with the aim being to keep it at a low and stable level. In addition to this there are concerns for keeping a stable level of output which should aim to keep it at a level around potential output and for general control of monetary aggregates such as money supply. With these objectives in place the Taylor rule can now be examined in how it allows central banks to follow a simple rule to meet its objective. The Taylor rule is a simple instrument rule which shows that interest rates should be determined by the inflation gap and the output gap as shown in this equation: (Walsh, 2003, p.546) The à ¯Ã‚ Ã‚ ¢ and à ¯Ã‚ Ã‚ § coefficients are both >0. This being added to the real interest rates leads to the Taylor principle which states that a deviation from the target rate of inflation should be met with a larger than one to one change in the nominal rate of interest. This is called the â€Å"Taylor Principle† and the empirical literature surrounding central bank behaviour aims to find evidence of the Taylor Principle adhered to by Central Banks. Bernanke (2004) describes the above equation as a simple feedback policy due to the central bank reacting to feedback from the economy on a number of variables that can be estimated at the time and don’t rely on forecasting. As the literature around the Taylor rule has grown so too has the variations of the model which have included both lagged variables and forecasting (Clarinda, et al.) It has also been adapted to provide guidelines for a variety of monetary policy regimes as Orphanides (2007, p.15) has pointed out two examples; one being a money growth regime and the other an inflation targeting regime. The Taylor Rule and the principle which follows on from it serve as a good starting point for monetary policy making due to its simplicity allowing a variety of variations of it to suit a variety of needs and thus serves a useful benchmark. Its simplicity provides a host of other benefits well. Firstly its ability to relate policy to the state of economy by showing how interest rates, inflation and output interact with each other it provides a good guideline for central banks to follow. In addition to if a central bank can commit to such a rule it will provide a baseline for expectations regarding future monetary policy for financial markets and other private agents. There are many criticisms of the Taylor Rule. Svensson (2003) and Woodford (2001) both imply that rules may be too simplistic to carry out the task of dictating monetary policy. Svensson (2003) also argues that it doesn’t contain enough economic variables to be useful. He mentions the exchange rate, terms of trade as well as others which may be of importance to a central bank in a small open economy. Thus he concludes that any policy using Taylor Instrument rules will be sub-optimal (Svensson, 2003, p.442). McCallum and Nelson rebut this by citing two models (Clarida et al. (2000) and McCallum Nelson (1999)) which are open-economy models which don’t require terms other than the interest rate, output and inflation rate. (McCallum Nelson, 2004, p.600) Tschandize et al. (2005) also points out that any recommendation based off of a formula is likely to ignore the judgment policymakers use in light of other developments not captured in the output gap or inflation behaviour. There are also practical problems with the Taylor rule though. Firstly the measures of both output and inflation can have a very different result depending on how they are measured (Yearly or Quarterly Data) and also due to measurement errors. (Orphanides, 2007) This could have a significant effect on parameters and lead to sub optimal policy making. Furthermore when there is deflation the Taylor rule if followed mechanically would demand a negative interest rate which is highly unlikely if not impossible due to the existence of a zero lower bound. Finally say if the inflation target was met and output was at its natural level then the rule dictates we set nominal rates at the real interest rate plus inflation. This presents numerous problems as there is extreme difficulty in measuring what is the natural long run rate of interest due to it being unobservable and having to be obtained implicitly. The Taylor rule is however generally held by all to be a good model considering its limited number of variables and serves as a good starting point for the oft complex task of creating monetary policy. Also if the Taylor rule is indeed followed as a rule many of the criticisms levelled against it are entirely valid, however if seen as a policy guideline rather than an iron clad law it is a lot more flexible and can instead inform policy makers rather than dictate them. Empirical Studies of the Taylor Rule Empirical studies tend to utilise rational expectations of forecasts, especially the model developed by Clarinda et al. this specification of the model is intuitively true as it would be rational to assume that central banks are forward looking in their policymaking due to the time lag between taking action and seeing that action having an effect it is better to take the action now for a forecast. In their study they find that the Taylor Principle held up well and you could accurately describe the policy undertaken by the Fed, Bundesbank and the Bank of Japan in the time frame studied. Clarinda et al go a step further and also include lagged variables of interest, regressions ran on interests rate with the coefficient on lagged inflation is both large and statistically significant implying serial correlation. For example Clarinda et al. find that with the fed two lagged variables of interest rates for the fed is both large and statistically significant. Some argue it implies that the fed is following an interest smoothing policy. This interest smoothing policy is intuitive for a number of reasons, for example central banks also use data from financial markets amongst others when deciding interest rates, and thus an interest smoothing policy would aim to not destabilize these other macroeconomic variables which would not be good for an economy’s wellbeing. This has been referred to as an illusion by Rudebusch (2002). He shows that if the Fed did adopt a gradual policy then it would be predictable but he argues that evidence from forward rates does not support this view. In addition to this Lansing shows econometrically why gradual smoothing appears. If the fed is using real time data to update its trend output each period then when the final data is produced due to the serial correlation between the real time errors will make it appear to be correlated with lagged interest rates. This creates the illusion of interest rate smoothing. More general points of criticism have been raised by many others (Perez(2001) Tschiadize et al. (2005) and Orphanides (2007). Perez (2001) argues argue that if we used real time data available to policymakers at the time we would find that the results do not hold up well and that in the period before the so called great deviation we would see that the Taylor rule was followed in the period of the great inflation (Perez, 2001). Orphanides (2007) argues that many studies have fallen into the trap of using revised ex-post data instead of the data available at the time, this error leads to results which provide no real insight into how decisions were made at the time. This point is also made by Tschandize (2005) Tschadize also points out that the structural change in an economy must be taken into account and thus it would be difficult to impose the same coefficients and targets on of one regime on another without accounting for structural changes. They elaborate by saying that while the structure of the economy may not change attitudes may change which may shift the result of the Taylor Rule equation due to different weights placed on the inflation variance and the output gap, and also a change in targets. Both of these would drastically change In addition to this many papers provide a counterfactual account of what should’ve been done. However with the benefit of hindsight and revised datasets it is very easy to say what should be done. Furthermore a study of this sort is of limited use as it is purely theoretical and is subject to the same limitations outlined above. They mention a 2003 study by Rogoff which shows that the smoothing of inflation may have occurred anyway due to favourable conditions in the macroeconomic environment, primarily globalization which put a downward pressure on prices due to increased competition from abroad so the evidence of Taylor Rules controlling inflation may be overstated. The empirical studies surrounding the Taylor rule have provided great insights into the conduct of monetary policy historically and have given insights into what works and has deepened our understanding of monetary policy. However there are many flaws in many of these studies which limits how many conclusions we can draw from them. Targeting Rules Could targeting rules provide a better alternative to an instrument rule? Svennson has been a strong advocate of targeting rules based on forecasting. One thing to note is that the Taylor rule is explicit whereas the model Svensson advocates is implicit in that inflation and output gaps matter but not because of themselves but in the way they affected the forecast for inflation. This particular model of optimal targeting relies heavily on developments made in consumption theory, Svensson argues that they are superior as they are structural, robust and compact. This model hinges on a very simple Euler Equation. This of course has come under criticism but it is irrefutable that it is compelling in its simplicity and its ability to distil the complexities of policymakers decision making into the very simple form of essentially MRS=MRT. The Euler condition is simply this: Et (Eq. 2 Svensson 2003, p.616) How does this relate to targeting rules? Targeting rules aim to minimise the loss between the marginal rate of substitution (MRS) between inflation and the output gap and the marginal rate of transformation between inflation and output is determined by the aggregate supply (AS) relationship between inflation and unemployment. Svensson (2005) notes that aggregate demand doesn’t determine the marginal rate of transformation (MRT), therefore the model is robust to changes in the AD relationship. This is an intuitive idea as policymakers have a preference over inflation and output just as a consumer has a preference over consumption today or tomorrow. Thus, a decision is made over how much output and inflation, which is dependent on the trade-off between them, which is given by the AS curve. So the principle of MRS=MRT can be applied to monetary policy. This principle is independent of any model and Svennson believes that this should drive a policymaker’s decision making not simply adhering to an instrument rule. Svennson (2005) outlines the central bank’s optimal targeting rule as: (Svensson, 2005, Eq.3) This rule is a structural model of monetary policy, in the same way that AS and AD are structural (and they are designed to capture price-setting and consumption choice respectively). As previously stated this essentially captures the equality MRS=MRT. MRS being given by the central banks preferences between inflation and output with à ¯Ã‚ Ã‚ ¬Ãƒ ¯Ã¢â€š ¬Ã‚  capturing the weight authorities place on output variability. The MRT being given by à ¯Ã‚ Ã‚ ¡x à ¯Ã¢â€š ¬Ã‚ ¬Ãƒ ¯Ã¢â€š ¬Ã‚  which is the slope of the short-run Philips curve which captures the trade-off between inflation and unemployment. Svennson (2005) says it is also robust to shocks and judgement since there is no variable in the rule to capture this. Finally he states that targeting rules are superior to instrument rules as they are more compact. This means that they can explain the same amount with less variables which can only be a good thing as it should lead to less errors. McCallum and Nelson (2004) argue that targeting rules are specific to a particular model. As they rely on assumptions of the dynamics of the models IS and Phillips curves amongst other structural equations. (2004, p.599) They criticise them as although they are optimal for a particular model they may well not be optimal in another model. In contrast they argue that instrument rules can be defined outside particular models and can be tested in other models, and that the best instrument rule over the range of models can be selected. They provide numerical examples in which the optimal rule in one model can give results in other models that are more than twice as bad as the optimum for that model (2004, p.599) They then run some simulations and conclude that there is little difference between the performance of instrument and targeting rules when a mistake is made regarding economic conditions. They argue that targeting rules are not superior to instrument rules in this respect. Svensson (2005) counters that if the error is not immediately realised, instrument rules can perform very badly. He also points out that whereas targeting rules are by definition optimal, varying the response coefficient in instrument rules finitely (rather than infinitely) can on some occasions only get close to optimality Targeting rules provide a good alternative to instrument rules and provide many benefits over instrument rules as shown above. That is not to say that it isn’t without its flaws but it does appear to more accurately model the behaviour of central banks as Conclusion Taylor rules and more generally instrument rules have been the focus of a great deal of economic research. The idea of a simple policy rule is an enticing one as it would be easy to commit to and would allow for an easy understanding of monetary policy. However the main issue is its simplicity as has been pointed out by many, central banks rely on all sorts of data when making monetary policy decisions. This thought process cannot be hoped to be captured in a simple instrument rule. It has found some success in empirical studies however with many showing that there is evidence of central banks making use of the Taylor rule and Principle but these findings should be taken with a pinch of salt as there are of course no certainties that central banks strictly followed a Taylor rule and also many critics have discredited some of the findings. However the results are still impressive considering the model has performed admirably in the years after it was first published in 1993 and still provides a compelling idea as to how monetary policy should be conducted and provides a reasonable explanation of central bank actions over the years. The development of optimal targeting rules has led to a compelling alternative to proposed instrument rules with its simplicity and strong micro foundations providing a model that holds up well to analysis. Indeed it is superior to instruments in a variety of ways due to its implicit nature and in the way it captures the principle that monetary policy is a case of getting MRS=MRT which is independent of any model and it leaves more scope for judgement to be used in how best to achieve this equilibrium. Of course it is not without its flaws such as its specificity to certain models and its inability to be used in other models. Unlike an instrument rule which is easy to apply and examine in a variety of models and the best rule can be selected. So the debate will continue and instrument rules and in particular the Taylor Rule are still relevant in the debate over the best way to conduct monetary policy due to its simplicity and it will serve as a useful guide for policymakers in the fu ture but the development of optimal targeting rules does provides a compelling alternative which is in my opinion a better model of monetary policy than any instrument rules as it more accurately captures the decisions facing policymakers due to its simplicity.