Tag Archives: fed fund rate

(Revised) Predicting the Effective Fed Fund Rate (3rd post)

In the last post, I talked about how market moves even without the intervention of fed trading desk. This might have gave you an illusion that OMO is less important in the determination of effective fed fund rate. However, it is the very commitment that trading desk will use OMO to keep effective rate around targets that leads to the market forming expectation, and ultimately actualize the expectation. From the graph below we can see that, in the long run, fed trading desk has done a great job in keeping the effr around the target . 

fredgraph

The average deviation of effr from target is about only 0.1 basis point. And there is evidence that the gap in between is shrinking overtime. This means that fed had become more and more apt at smoothing macro economic shocks. It could also mean that the market have been getting better and better at predicting the move of trading desk, out of the expectation that it will always follow the reaction function.

So there is obviously a gain for the forecaster who wants to forecast effective fed fund rate in the longer horizon. It is difficult for forecaster to use the traditional approach where supply and demand of fed fund is estimated and then the equilibrium fed fund rate is calculated. It is especially tenuous for forecaster to estimate the demand for federal funds. As I discussed earlier, demand for fed balance is now largely driven by loaning opportunities in hundreds of other markets, rather than driven by the need to fulfill legal reserve requirement.  There is an interesting research paper about topology of fed fund market, where it shows that larger banks usually plays a role as fund buyers and smaller banks fund suppliers. Larger banks tend to have more credits and more lending opportunities. Those small banks fund larger banks by selling extra reserves. Thus it became harder and harder to trace out exactly how much each bank need the fed balance by simply looking at how much they need to hold at the Fed.

More importantly, since effective fed fund rate is moving closely around the target and policy makers set the targets manually, it is obvious that the fed has been playing more important role in determining the fed fund rate. Estimating demand is less important. The Federal Reserve Bank set the target rate to make sure sufficient credits are available for the economy, and they adjust the target rate according to a large set of other macro-economic indicators.  The fed trading desk make sure effr does not go astray the target rate.

The fact that effr does not deviate fed target rate for too long and by too much is convenient for forecasting fed fund rate in the longer horizon. Now we can focus on forecasting the fed fund target rate. Fed fund target rate is determined by FOMC who keep track of bunch of macroeconomic indicators.  If we can approximate the indicators used by them in making decision of next period fed fund target rate, we can also forecast the movement of fed fund target rate. Forecasting effective fed fund rate would, therefore, be roughly the same as forecasting fed fund target rate.

(Revised) Predicting Fed fund rate (2nd post)

In my last post, I talked about how effective fed fund rates are influenced by two market forces: the banks and trading desk that takes directives from FOMC. I also talked about the difficulties to forecast effective fed fund rate. In this post I will continue to talk about how to predict effective fed fund rates.

We all know to this point that effective fed fund rate is determined by the market. But, fed trading desk has done a great job in counteracting deviation of effective fed fund rate from target rates. The market has observed this pattern, too. Therefore, expectations of actions of fed trading desk have been widely formed, guiding the banks’ movement.

According to John Taylor, when FOMC announced the new target rate, the market anticipate that trading desk’s will perform OMO, which will bring effr to its targets.  Therefore the banks move even before OMO, bringing effr to its new target level. Rational expectation plays a part because market agencies know the reaction of fed trading desk through their day to day interaction, and they take this reaction into account when they choose how much funds to buy. How much fed fund they want to buy depends on effr today and expectation of effrs in the future. If they predict that fed fund rate will rise tomorrow, they will buy more reserve today, and the effect of which is that the fed fund rate will rise today. Same logic apply: If they expect the fed trading desk will perform OMO that bring up fed fund rate tomorrow, they will demand more fund today.

To be more specific, I will give an example. Suppose that the Fed announced today that the target would rise by 0.5 basis points. What will trading desk do? Unfortunately, Trading Desk cannot do anything that day, because the Desk typically enter into the market at 9:30 am before the day of trading start and before the FOMC announced the new targets. It is important to know that the trading desk action is based on last period’s information.

On the other hand, traders in the market can react immediately to the new targets. They know the Trading Desk will perform OMO tomorrow to bring effr up to new target. Knowing for sure that the future fed fund rate will rise, traders will try to buy more funds today to avoid higher expense tomorrow. This increases the total demand in the market, which eventually bringing effr up almost within a few days to the new target rate. 

The fact that effr does not deviate fed target rate for too long and by too much is convenient for forecasting fed fund rate in the longer horizon. I will continue to talk about how to incorporate this information into forecasting model in the next post.

(Revised) Predicting the fed fund rate (1st post)

Since 2008, federal fund rates have been targeted within a range of 0%-0. 25%. Therefore, prediction of this rate seems to be less attractive at this particular moment. However, given the important status of fed fund rate in financial world and the fact that it will return to positive in the future, it would be interesting to see how economists forecast fed fund rates.

To make forecasting more accurate, it is important to distinguish factors that driving the federal fund rates. Effective fed fund rate is determined by the federal fund market, where Financial institutions trade federal fund with each other, usually over night, on an un-collateralized basis. There are two major players in this market: traders employed by financial institutions, and the Trading Desk who takes command from Federal Open Market Committee (FOMC). These market forces are what ultimately influencing the effective fed fund rate.

On the one hand, banks with surplus balances direct their traders to bid a higher price and banks with a deficit bid a lower price.  Brokers find the match of those bids to facilitate the transactions. At the end of each day, a weighted average of all settlement prices is the effective federal funds rate. On the other hand, fed trading desk intervene in the fed fund market through buying and selling treasury notes in the Treasury markets (not the fed fund markets), so as to influence the reserve balance held at banks and keep the effective fed fund rate moving around the federal fund target rate. This is known as Open Market Operation (OMO). The federal funds target rate is determined by a meeting of the members of the FOMC, normally occurs eight times a year about seven weeks apart.

Beginning in 1994, the FOMC began issuing changes about the target fed fund rate explicitly. This greatly affect the Fed fund market behavior, and this effect has been discussed in Expectation, Open Market Operations and Changes in the Federal Fund Rates, a paper by John B. Taylor at ST. louis FED. According John Taylor, the effective fed fund rate responds to deviations of it from the target quickly, even without the Trading desk’s intervention (or OMO). This model might be expected to forecast well at short horizons but less well at longer horizons.

It would be tempting to use the supply and demand model to predict the effective fed fund rate. However, this is a less effective method because nowadays banks holding reserves not solely out of the need to meet legal requirement. Modern technology facilitates the emergence of “sweep” account.Banks “sweep” their customers accounts from those with reserve requirement to those without reserve requirements. Typically, the bank can sweep the checking account  into more lucrative Money Market Mutual Funds.  Another evidence that fed fund markets plays more than a role of helping member banks meet reserve requirements is the fact that, the flow of trade in the federal funds markets is more than ten times greater than the stock of fed balance. Banks typically buy funds in the market to make loans in other markets. Fed funds purchases are a source of funds for other loans creates a connection between the federal funds rate and the other loan rates.

Thus, forecasting fed fund rate is not as intuitive as it seems.

Luckily, if one closely observes the activities of market participants, predicting future fed fund rate is not impossible. I will continue talk more about how to predict fed fund rate in the next post

 

How to predict fed fund target rates

The importance of fed targeting fed fund rate is to foster price stability and full employment by systematically reducing uncertainty and increasing the credibility of future actions by the central bank. Since 1994, the Federal Open Market Committee began issue a statement whenever it increased or decreased its target for the federal funds rates.

Given the key role fed fund target rate plays in the economic and financial system, the determination of it has been given great deliberation by the FOMC during their meetings held approximately every six weeks. It is predictable that the target rate is the outcome of a complicated decision-making process. Numerous economic indicators are closely monitored by the FOMC, in order to determine the most appropriate course of action.

In economy theory, we know that the fed target rate is roughly determined by the deviation of actual inflation rate from target inflation rates and of actual GDP from potential GDP, this is known as the Taylor Rule: 

i_t = \pi_t + r_t^* + a_\pi  ( \pi_t - \pi_t^* )  + a_y ( y_t - \bar y_t ).

But apparently GDP and inflation are not the only elements factored into the complicated decision function by the FOMC. According to a research paper: Bayesian Forcasting of the Federal Target Rate Decision done by Sjoerd van den Hauwe, Richard Paap and Dick van Dijk, the minutes of FOMC meetings indicating that a large number of other economic variables, reflecting developments in the labor market, housing market, and financial markets, also play a substantial role in the considerations.

For example, the following variables, among others, are explicitly mentioned in the minutes of the FOMC meeting held on December 15-16, 2008: private payrolls, new claims for unemployment insurance, industrial production, factory utilization rate, consumer spending, consumer sentiment, new building permits, house prices, vacancy rates, etc.

The paper aim to assess which macroeconomic and financial variables are most informative for the FOMC’s federal funds target rate decisions from a forecasting perspective. They analyzed 157 target rate decisions made during the period January 1990 – June 2008, and consider a set of 33 possible predictors. The paper successfully incorporate these 33 predictors into the model and 82% of the FOMC decision of whether to change or maintain the targets are predicted correctly.

The variables in this set are classified into three categories. First set include recent releases of macro variables such as output, employment and inflation, reflecting the fact that these are most directly related to the Federal Reserve’s monetary policy goals. Second set include several other macroeconomic and financial variables,  most of which are established leading indicators, providing signals about future economic developments that are potentially useful for predicting FOMC decisions. This set also included forward-looking nature of asset prices such as stock prices and interest rates, which has been shown to result in predictive ability for macro variables such as output and inflation. Third set of variables include survey measures of consumer confidence and expectations as well as professional forecasts for inflation, output and interest rates. This is motivated by the results in Ang et al. (2007), who demonstrate that survey-based measures and forecasts outperform macro variables and asset prices in forecasting inflation.

I think if would be very useful to include these variables into my forecasting model for monthly fed fund rate. Although effective fed fund rate deviate from fed fund target rate by a certain amount, for monthly data that deviation would be minute. As I discussed in my last post, the effective fed fund rate quickly respond to the target rate change, and fed trading desk has successfully kept effective rates close to its targets. Therefore, for short frequency forecasting, forecasting the decision of FOMC would be roughly equivalent to forecasting the effective fed fund rate.

 

Fed fund rate: not as simple as it seems. (4th post)

In my last post, I talked about the fact that effective fed fund rate move even without OMO actually taking place, bringing the fed fund rate towards its new targets. This might have gave you an illusion that the market is moving without the intervention of fed. However, it is the very commitment that trading desk will use OMO to keep effective rate around targets that leads to the market forming expectation, and ultimately actualize the expectation. That is to say, the fed stir up market sentiments to cooperatively bring effective fed fund rate towards target rate.

From the graph below we can see that, in the long run, fed trading desk has done a great job in keeping the effr around the target (click on the graph to view the original image). fredgraph

The average deviation of effr from target is about only 0.1 basis point. And there is evidence that the gap in between is shrinking overtime, meaning fed had become more and more apt at smoothing macro economic shocks. It could also mean that the market have been getting better and better at predicting the move of trading desk,out of the expectation that it will always follow the reaction function.

So there is obviously a gain for the forecaster who want to forecast effective fed fund rate in the longer horizon. It is difficult for forecaster to use the traditional approach where supply and demand of fed funds are estimated and then the equilibrium fed fund rate is calculated. It is especially a tenuous task for forecaster to estimate the demand for federal funds. As I discussed in my second post, demand for fed balance is now largely driven by loaning opportunities in hundreds of other markets, rather than driven by the need to fulfill legal reserve requirement.  There is an interesting research paper about topology of  fed fund market, where it shows that larger banks usually play a role as fund buyers and smaller banks fund suppliers. Larger banks tend to have more credits and more lending opportunities. Those small banks fund larger banks by selling their extra reserves. Thus it became harder and harder to trace out exactly how much each bank need the fed balance by simply looking at how much they need to hold at the Fed.

Fortunately, the Federal reserve bank economists had done the job for us. Fed had already estimated the total liquidity and corresponding amount of fed balance needed by banking system, and they set the target rate to make sure sufficient but not overwhelming credits are available for the economy. To make sure of this, the effr must not go astray the target rate. The fed trading desk make sure of this by performing OMO everyday in the market. The fact that effr does not deviate fed target rate for too long and by too much is convenient, since now we can just forecast the fed fund target rate. As we all know,  fed fund target rate is determined by fed official who keep track of bunch of macroeconomic indicators.  If we can approximate the indicators used by Fed in making decision of next period fed fund target rate, we can also forecast the movement of fed fund target rate. Forecasting effective fed fund rate would, therefore,  be roughly the same as forecasting fed fund target rate.

Fed fund rate: not as simple as it seems. (3rd post)

As I discussed in my last post, effective fed fund rate responded to announcement of target rate changes immediately even before open market operation. The most important reason for this phenomenon being that FOMC communicates the new target rate to the market more explicitly. Apparently, expectation about fed behavior has widely formed in the fed fund market, but how does this expectation formed ?

To understand how rational expectation plays a part, it is important to understand that fed fund market know the reaction of fed trading desk towards a change of target rate, and they take this reaction into account when they choose how much funds to buy. So what is the reaction function of the trading desk and what is the trader’s demand for fed balance? In Taylor’s paper, the reaction function is:

D014AAD7-942B-420C-A40D-36F083668E1E

where bt stands for the supply of fed balance at time t. r superscript e stands for effective fed fund rate and r superscript t stands for fed fund rate targets. This function means that the supply of fed balance the trading desk trying to manipulate today depends on fed balance yesterday and the deviation of effective fed fund rate from target yesterday. The second part reflect the truth that FOMC directs the trading desk to but and sell securities so that conditions in the federal funds market are consistent with and average federal funds rate near the targets.

This reaction function is known by all the fed fund traders through their day to day interaction with the trading desk.

What does fed fund trader trying to achieve when they participate in the market? They try maximize the banks profit by getting the cheapest fed funds! And they no only try to get the cheapest deal today, but also tomorrow. So if they predict that fed fund rate will change tomorrow, they will react today. For example, if they predict that fed fund rate will rise tomorrow, they will buy more reserve today, and the effect which is that the fed fund rate will rise today!

the demand for fed fund by each trader at a particular day is characterize as:

511D7317-1FC0-4450-8AAA-5027D37D6677

if we ignore alpha and gamma for now, we can interpret the function as the demand for fed fund depends on the difference between fed fund rate today and the expectation of fed fund rate tomorrow. The expectation of future rate is a forecast by fed fund traders incorporating all the information available today and the rational expectation of fed trading desk’s reaction tomorrow. If they expect the fed trading desk will perform OMO that bring up fed fund rate tomorrow, they will demand more fund today.

Given our two reaction function above, we have enough ingredients to tell a story of how fed fund rate adjust to new target rate even without OMO.

suppose that the fed announced today that the target will rise by 0.5 basis point. What will trading desk do? Unfortunately, trading desk cannot do anything that day because its reaction function depends on yesterdays information, before the FOMC announced the new targets. This is especially credible given the fact that the trading desk typically enter into the market at 9:30 am before the day of trading start. Again, the trading desk increasing or decreasing of the fed balance is based on estimation of last period’s information.

On the other hand, traders in the market know all the information, both the fact that target is changing and that trading desk perform OMO tomorrow to bring fed fund rate up to new target. The trader formed rational expectation given its own demand function and reaction function of the desk. Knowing for sure that the future fed fund rate will rise, traders will try to buy more funds today to avoid higher expense tomorrow. This increase the total demand in the market and eventually, the market move before the trading desk perform OMO the next day, bringing effr jump a huge step, nearly reaching the new target rate. 

 

Fed fund rate: not as simple as it seems

Given the fact that I am writing a term paper for effective fed fund rate prediction, I would like to write several posts on this topic. This is first post of the stream.

First off, It is important to distinguish between the two term, effective fed fund rate and fed fund rate targets. 

Effective fed fund rate is determined by the markets, that is, the federal fund market, where Financial institutions trade federal fund with each other, usually over night, on an uncollateralized basis. Federal funds are mandatory minimum reserve balance held at Fed reserve. For banks to meet the requirement on a daily basis, existence of such a market is essential. There are three players in this market: 1)Traders employed by financial institutions, 2)brokers, and 3) Fed trading desk.  So how this market works? Banks with surplus balances direct their traders to quote a highest price (interest rate) and banks with a deficit quote a lower price.  Brokers find the match of those bids and quotes and earn commissions once a deal is achieved. Fed trading desk intervene the fed fund market by operating on the Treasury markets( note, not the fed fund markets), a behavior known as Open market operation. At the end of each day, the interest rate for each successful transactions between banks in the fed fund market is recorded and the weighted average of those rates is the effective federal funds rate.

The federal funds target rate is determined by a meeting of the members of the Federal Open Market Committee which normally occurs eight times a year about seven weeks apart. Began from 1994, the FOMC began issuing a new statement about the target fed fund rate explicitly. This greatly affect the Fed fund market behavior, and this effect has been discussed in Expectation, Open Market Operations and Changes in the Federal Fund Rates, a paper by John B. Taylor at ST. louis FED. According John Taylor, the effective fed fund rate responds to deviations of it from the target quickly, even without the Trading desk’s intervention (or OMO) . This model might be expected to forecast well at short horizons but less well at longer horizons. There are other reasons discussed in the paper of why fed fund rate seems to react less and less dependent on the open market operations. One of the reasons is the lagged reserve requirement system, and the other reason is that technology changes reduced the need of banks to hold reserve for legal purpose. I will talk more about this three reasons in my next post.