The rationality evaluation in chapter 2 examines the unbiasedness and efficiency of the interestrateforecasts from the WSJ survey. During the non-crisis time period, almost all the funds rateforecasts are not significantly biased. But during the financial crisis, most of the professional forecasters missed the declines of the actual funds rate. The majority of the forecasters make note yield forecasts significantly higher than the actual note yield. The inefficiencies in the forecasts mainly come from not efficiently using the past changes of the actual funds rate or the actual note yield. Some degree of rigidity are found in the funds rateforecasts from the WSJ survey and the T-bill rateforecasts from the SPF. They possibly are caused by imperfect information.
Applying the problem of value of information raised by Morris and Shin (2005) into the specific form of transparency, we can think of a ‘self-fulfilling mechanism’ of central bank’s and private’s information, keeping in mind the dual role of central bank as shaper and observer of the market. That is, when private information is very accurate, and if the precision of central bank signal (in the form of central bank’s interestrateforecasts) increases, at a level that crowds out private information, agents take actions basing on the over-weighted central bank information. Then the agents’ actions bear less information value, or loosely reflect the true conditions of the economy. Now, at the next period, forecasts becomes less accurate, since it is formed basing on the signals of prices which are now less informative, until it returns to a level that does not crowd out private information any more. From that period, private information is not under-weighted and investors’ actions be- come more informative. At the next period central bank forecasts gaining more informativeness from market signals becomes more precise back, and so on until its precision reaches the level that crowds out private informa- tion again, and so on. We call this a ‘self-fulfilling mechanism’ of central bank’s and private information which will maintain the relationship between these types of information within a corridor in form like a sin-shape graph. However, because learning is possible for both central bank and financial market participants, this shape might have decreasing amplitudes overtime and there will be no longer difference between central bank’s and private sector’s forecasts in the infinity horizon (someday, all available information become common knowledge).
and the Riksbank. They conclude that the conditional forecasts published by these institutions were of little value to market participants. Instead, they argue, central banks should show un- conditional forecasts, based on the most likely path of interest rates. Ferrero and Secchi (2007) review quantitative and qualitative interestrateforecasts of four central banks and conclude that their publication improves the ability of market operators to predict monetary policy decisions. Euseppi and Preston (2007) show that when the central bank does not have full knowledge about the economy communicating details about monetary policy rules helps restore stability. Rudebush and Williams (2006) use a standard New Keynesian model with learning to show that publishing the interestrate path lowers the variability of output and inflation. The gains increase with the difficulty to infer the objectives of monetary authorities. To our knowledge no study attempted to compare the gains from publishing macroeconomic projections and the interestrate path. Our paper tries to fill this gap.
There are a number of interesting features of the statistics reported in the tables. As expected, given the uncertainties associated with the exchange rate, the proportion of wealth in foreign assets falls as the risk aversion parameter rises. So, for example, if we simply average the figures in the columns of Table 6a, the share allocated to UK assets falls from 43% when A = 2 to 33% when A = 5 and to 23% when A = 10. The share of UK assets based on forecasts from the equal-weight average model is 31% when averaged over the various investment horizons and for A = 5. Here, where the AIC weights were distributed relatively widely across the models, this is reasonably close to the 26% suggested by the AIC-weighted average model. 17 The average results accommodate considerable heterogeneity in outcome across the various models, however, as shown in the three examples provided in the table. Hence, again taking the averages across the investment horizons and with A = 5, model M EV suggests a holding of 45%
Herbst and Schorfheide (2011) also use PITs to evaluate density forecasts of a small New Keyne- sian model and the Smets & Wouters model. They find that for the Smets & Wouters model the density forecasts for output growth are too wide. The interestrateforecasts of both models are skewed as the federal funds rate was lower than predicted over the evaluation sample. The inflation forecasts of both models and the output forecasts of the small New Keynesian model perform well. The differences to my results might be due to differences in the sample and the evaluation data. Their estimation sample starts in 1984, while I use a rolling window of the most recent 80 quarterly observations. Their evalu- ation sample ranges from 1997 to 2004, while my evaluation sample starts in 1984 and ends in 2000. Furthermore, they deviate from Smets and Wouters (2007) by calibrating some parameters that have been estimated by Smets & Wouters. Herbst & Schorfheide use final revised data for the evaluation of the forecasts, while I use data that includes initial revisions only. At least in the first half of my estimation and evaluation samples I estimate the models on relatively volatile data, but evaluate the forecasts on data that belong to the great moderation. This puts the DSGE models to a very hard test and might partly explain the overestimation of actual uncertainty. The evaluation sample of Herbst and Schorfheide (2011) is relatively short and includes the high productivity growth period of the late 1990s and only one very mild recession in 2001. They estimate the models on data with low volatility and also evaluate the models on data of low volatility. The inclusion of the recent financial crisis could potentially lead to a reduction in the difference of predicted and actual uncertainty relative to my cur- rent results (at least at the left tail of the forecast distribution), but to an underestimation of actual uncertainty in the Herbst & Schorfheide case. 23 I leave it for future research to assess the sensitivity of DSGE models’ density forecast performance to the estimation and evaluation sample.
However, many economists take a different view, such as Cecchetti et al. (2000), Borio and Lowe (2002), Bordo and Jeanne (2002), Platen and Semmler (2009), and Smets (2014). They argue that low and stable inflation is not a guarantee of financial stability. Therefore, central banks should systematically incorporate asset prices into their policy-making pro- cesses to improve the effectiveness of monetary policy. As the collapse of an asset price bubble may lead to a financial and economic crisis, they urge central banks to intervene pre-emptively by raising interest rates and hence preventing the build-up of an asset price bubble (‘leaning against the wind’ policy). The authors point out that the associated devia- tions from the Taylor rule should prevent episodes of deflation and recession which may occur after the bursting of an asset price bubble. Juselius et al. (2016) presented an aug- mented Taylor rule by adding a financial cycle variable. This enables to take financial devel- opments systematically into account. However, the alleged unavailability of timely warning indicators is not per se a hindrance to implementing a ‘leaning against the wind’ policy as Alessi and Detken (2009) found well performing early warning indicators. By contrast, Bernanke and Gertler (2001) argue that the ‘leaning against the wind’ policy only performs well if central banks know that the boom is driven by ‘irrational exuberance’ and a collapse of asset prices is imminent. These are both highly unlikely assumptions. Additionally, Blot et al. (2017) suggest that movements in asset prices driven by fundamentals should be dis- entangled from movements resulting from the speculative component. Furthermore, Galí (2014) claims that increasing interest rates as a response to asset price bubbles cause op- posing effects on fundamental and non-fundamental components of asset bubbles. Thus, when it comes to rational asset price bubbles, the ‘leaning against the wind policy’ may be counterproductive and might cause higher bubble fluctuations as well as adverse effects on the economy. Berlemann and Freese (2013) mention that tightening of the monetary policy could deflate house and flat prices while increasing rental prices and thus inflation.
In the previous chapter we specified coefficients of the Chaos Models, without loss of generality, by applying the exponential polynomial family. This specification allows us to compute both initial yield and volatilities at the same time. However, we start our calibration by looking at only initial yield curves. Here, our main concern in this chapter is to check if our specification of the chaos coefficients allows to fit well into the initial yield curves. As seen in , the Nelson-Siegel Form () and the Svensson Form () are the ones that most central banks apply, with the exception of those in Japan, UK and USA which apply Smoothing splines. These forms may be regarded as a special case of the general parametric form suggested by Bj¨ ork and Christensen in . Unfortunately this model has the shortcoming that it allows negative interest rates. We compare initial curve fitting ability of the Chaos Models with those parametric forms and also among different chaos orders by using data from the UK bond market. We show that the proposed model attains just as good a fitting to yields as the Svensson Form does, while also keeping the interestrate positivity condition.
This table presents the panel estimation results for the regression which evaluates the bank holding companies’ (BHC) interestrate risk with respect to the maturity of securitized assets over the 2001 to 2009 period. The dependent variable is the absolute value of the coefficient measuring the sensitivity of BHC’s i equity returns to unanticipated changes in the slope of the US sovereign zero-coupon yield curve at year t. This coefficient is estimated from a four factor GARCH market model. Specifically, for each bank-year, I run a four-factor time series regression of BHC weekly returns on the market returns (MRK), and unanticipated changes in zero-coupon yield curve level (LEV), slope (SLO), and curvature (CUR). The estimation requires at least 30 weekly return observations for each bank-year. The corresponding US zero-coupon yield curve level, slope, and curvature are estimated using Diebold and Lee (2006) parameterization of Nelson and Siegel (1987) model. The unanticipated changes in the yield curve factors at time t are calculated as the difference between the actual changes in these factors and ones forecasted via an appropriate specification of the autoregressive moving average (ARMA) model. A bank holding company is defined as securitizer (columns 1 to 4) if it reports at least one securitization transaction over the analysed period in Schedule HC-S of the Federal Reserve System’s FY-9C filings. The explanatory variables on the right-hand side are as follows: TSEC is the outstanding principle balance of assets securitized or sold measured as the proportion of total assets; the outstanding balance of securitized long-, medium-, and short-term loans are LT_SEC, MT_SEC, and ST_SEC respectively; the asset growth rate (AGR); equity capital (CAP) calculated as the ratio of BHC’s book value of equity capital to its total assets; H_NITR(H_NOIR) is the Herfindahl-Hirschman (non)interest revenue concentration index calculated on the basis of twelve (eight) part breakdown of the(non)interest income;the proportion of total assets that are liquid (LATA); H_LOAN is the Herfindahl-Hirschman loan concentration index computed considering five loan categories; GAP is the balance sheet maturity gap calculated as the difference between interest-earning assets and interest-bearing liabilities maturing or being repriced within one year, scaled by the bank’s total assets; NECP is the net credit protection (protection bought minus sold) purchased by a bank; the ratio of non- performing loans to total loans is NPL; ROID is the measure of bank revenue diversification; and return on assets is represented by ROA. The regression also includes year- and state-dummies which are not reported. Heteroskedasticity and autocorrelation consistent t-values based on White’s robust standard error are reported in italics. ***, **, and * represent significance at the 1%, 5%, and 10% levels, respectively.
suggest some solutions. In another study, Buston (2016) found that banks with active risk management were less likely to become insolvent during the crisis of 2007–2008, even though their balance sheets displayed higher risk-taking. Hryckiewicz and Kozłowski (2017) analyzed the banking sector risk structure during the 2007-2008 financial crisis and demonstrated that during the financial crisis, the funding structure was responsible for the systemic effect of the financial crisis. Further, they demonstrated that countries with systemically important banks that rely on investment activities experience a greater, but more short-lived decline in GDP, when compared with countries that have predominantly traditional banking activities. Williams (2016) also found that the 2007-2008 financial crisis changed some aspects of the relationship between bank risk and revenue composition. Non-interest income is generally found to be risk increasing, but some types of non-interest income are risk reducing when bank specialization effects are considered.
rate differential on its change depends on its own size. Arbitrage activities will occur only if the lagged differential is large enough. Otherwise the arbitrage activities may stop because of the transactions costs. This model is from Mancuso, Goodwin, and Grennes . The EDF in this table generally supports the nonlinearity relationship in this model. We find EDFs from the U.K. and Switzerland are close to one and the corresponding smoothing parameter is relatively large. From Figure 1.7 we can find the marginal response from these two countries are close to a straight line. Compared with the result from linear version models (red lines), nonlinearity features in the model seems not to affect the result at all. However, at the same time, strong nonlinearity relationships are found in the results from Japan and the Eurozone. The EDF is close to nine and smoothing parameter is close to zero. Comparing the blue lines with red lines in the figures, we find the marginal effects of lagged differentials indicate large negative outliers lead to a significant positive adjustment to the differential, which is strong evidence of the existence of the nonlinearity relationship in dynamic linkages of real interest rates in international markets.
Starting with the mean, the Japanese yen/U.S. dollar exchange rate with a mean value of 105 reveals Japan as the economy with the weakest currency relative to the U.S. dollar, when compared to the exchange rates of the other G7 countries. The average interestrate for the period under consideration is 2% for Canada and the United Kingdom, which is the highest, and Japan is the country with the lowest interest rates. The inference from the standard deviation also is that Japan is the country with the most volatile exchange rate, while the United Kingdom has the least volatile exchange rate. However, as expected of developed economies, the standard deviation for the interestrate seems to be reasonable for both the euro and non-euro G7 countries. With respect to the distributions, we find evidence of non-zero skewness for all the series and across all the G7 countries. The kurtosis statistic is, however, mostly platykurtic in a number of countries, except in a few instances in the case of Japan, Canada, and the euro area. These differences could support our assumption of different responses of exchange rates to changes in the interestrate differential across countries.
The possibility that an aversion for policy reversals would lead the central banks to be conservative is pointed out by Lowe and Ellis (1997), while I introduce the central banks’ reversal aversion in a more formal way. I consider a situation in which the central bank faces an irreversibility constraint that prohibits the policy rate to move in opposite directions over the two consecutive periods. This can be interpreted as a special case of the situation in which the central bank bears the cost of policy-rate reversals without any restriction on the control space. Thus, as noted by Lowe and Ellis (1997), this discussion is similar to the irreversibility of investment (e.g., Dixit and Pindyck, 1994). The introduction of irreversibility makes the optimal policy less aggressive than would be attained under no constraint. This is simply because a larger policy shift would increase the probability that the policy rate must be reversed in the next period. In other words, there arises an option value to wait as is the case with investment irreversibility.
are much more volatile. The SHIBOR rate remains the lowest among these three money market rates, but still far above the lower bound (the interestrate on the excess reserves), which is 0.72% throughout the period and not presented in the figure. The money market rates, in all three measures, were low before 2017 and lying below the upper bound of the SLF rate. Afterwards, the market interest rates, especially two repo rates, went up in March 2017. Though the SLF was also adjusted upwards, both the repo rate and the interbank pledged repo rate happened to be very volatile and broke through the upper bound of the corridor on many days. It seems that since 2017, these two money market interest rates have not been effectively confined to the corridor. Only the reference money market rate, the SHIBOR, has moved within the corridor. The possible reason for this ineffective upper bound might be due to the fact that the banks cannot borrow from the PBC as much as they wanted with the SLF, which in turn has resulted in a market interestrate above the SLF rate on those days of high demand for liquidity.
Over the past two decades, China has adopted numerous policy changes to advance its financial market. Interestrate liberalization is one of the most important changes in this process. The Chinese interestrate market transformed from a fully controlled market to a dual-track interestrate market, under which banks and capital markets work together on the monetary resource allocation. However, regulatory controls over interest rates have not yet been implemented. A distinct market distortion along with quantitative controls also exists on credits. In this paper, we consider both international experiences and Chinese national situation when analyzing the choice of a benchmark rate for China. Although China currently does not have any interestrate which is perfect for benchmark rate, China can adopt a benchmark rate similar to LIBOR or the US Federal funds rate.
Machaj abstracts from the output mix in his examples, and thus we cannot be sure whether any of them represent a lengthened structure of production, notwithstanding the appearance that this has happened by focusing on the temporal aspect of production. In conclusion, changes to savings preferences alter the “length” of the structure of production, which is reflected in the interestrate. In the unhampered economy, the interestrate does not change the structure of production but rather it is through preference shifts between present and future goods on the structure of production in conjunction with the credit market that the interestrate obtains. Of course, the role of the production structure in determining the rate of interest on the loan market has been discussed already and at length in Rothbard (1962: chap. 6 and esp. 378).
Licensed under Creative Common Page 343 On a similar study, Furman & Stiglitz (2006) from their study in nine developing nations form 1992-1998, shows that an increase in interest rates can result in a deprecation of exchange rates. However, the finding of this study is more suitable in countries with low inflation than in countries with high inflation. In another viewpoint, Purnomo (2017) took a different dimension on the effect of inflation on the currency exchange rate in an Islamic finance point of view using an interactive analysis techniques. The Author argued that, the decline of inflation in Islamic economy was due to slow economy growth influenced by the weakening of world economy and thus affected the unpredictable exchange rate of the Islamic world. exchange rate can also be affected by other factors such as income level, government control and speculations on future exchange rates (Tafa, 2015).
I create a model where private banks face adjustment costs in nominal interest rates. The model’s inflation responds to interestrate changes (both nominal and real) by mov- ing in the opposite direction. That response justifies the Taylor rule and explains, through credit conditions, the procyclicality of inflation. The model permits the analysis of dif- ferent types of monetary policy using a variable inflation target. I use this feature to simulate different policies and compare them to interestrate data from the last century. The interestrate rigidity model leads to credit-conditions-driven inflation, which I be- lieve is more realistic than competing models of inflation.
Recall that when the money supply is specified the equilibrium price vector and the equilibrium wheat output in this economy depend only on the interestrate. If the interestrate is set to an exogenous level without adjusting the money supply (as the monetary policy discussed in ), the economy may converge to a specified equilibrium. For instance, if the money-owner can control the interestrate, the interestrate will be set to 1 to maximize her utility. If the laborer can control the interestrate, the interestrate will be set to 0. Fig. 2 shows the dynamics of the economy with an exogenous interestrate 1, wherein the money supply is 100 dollars all the time.
This value can be divided over the estimated number of periods, sub-periods, envisioned for a low annual or monthly payment. The compound interest formula future value would bring hardship to any middle to upper-class family, on monthly basis. On the bases of the above calculation it is appropriate to compare the two methods for a large loan over different loan periods to emphasize the disparity in savings between the two methodologies. Such would amplify the great value of the new technique in future transactions. Let us assume that the amount borrowed is $200,000. This is close to the average amount borrowed for starting a business, buying a house, or making some expensive purchases. Table 3 provides us with ample data comparing the compound interest formula to the new non-compound Interest Formula. Looking at the data on Table 3 we can easily see the difference between future values of Equation 3 and those of Equation (13). The earnings may be much lower for the non-compound interest formula compared to those of the compound interest formula, however, it is the economic situation, nationally and internationally that may determine which is more appropriate.
Our results indicate the largest di¤erences across models and methods are for CA. Graphs of the daily returns in the out-of-sample period (July 4, 2002 to October 27, 2003) along with 5% VaR forecasts from the AR and HAR models (computed assuming a normal distribution) shed some light on these …ndings. Figure 3 shows that after April 2003, the frequency of large negative returns increased. Before this point, there is almost no di¤erence across models and methods in the VaR forecasts. After April 2003 it is apparent from the …gure that there are marked di¤erences between the HAR and AR model VaR forecasts. Figure 4 shows the 1-step ahead forecasts of realized volatility (and outturns) for all …ve countries. From this …gure, it is clear that the good performance of the HAR model VaR forecasts for CA stems from the superior performance of the HAR volatility forecasts over this period. The HAR forecasts are better able to capture the general upturn in volatility relative to either the AR or MIDAS models. It is also apparent from …gure 3 that currencies other than CA do not show such a clear level shift, or such a clear distinction between the volatility forecasts of the models.