The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
1 Hello, I’m Ben Chabot – Financial Economist at the Federal Reserve Bank of Chicago. Today I’m going to talk about the overnight money market. Understanding overnight money market dynamics and participants is crucial to the effective implementation of monetary policy. Even more so during the policy normalization process. I will start by introducing various safe short-term investments available to investors with cash management needs. Then I will briefly describe the traditional banking system and elaborate on what we call the shadow banking system. We will talk about how the Fed interacts with the traditional banking system to target the overnight federal funds rate and how the Fed can use its new tools, that is, the interest on excess reserves and reverse repurchase agreements, to influence overnight interest rates in both the traditional and the shadow banking systems. Many investors want to hold cash or cash-like overnight investments. To this end, we refer to them as cash pools. This demand is primarily from global corporations, large institutional investors, large asset managers, security lenders, government-sponsored enterprises, foreign central banks, and state and local governments—all of which need cash for payment and liquidity purposes. When we say “cash-like” investments we mean investments that match the safety and utility of holding cash. Cash is risk-free. The nominal value of cash is invariant to changes in interest rates, so a cash investment is worth the same amount tomorrow as it is today. Not surprisingly, many investors wish to hold cash-like investments to satisfy their payment and liquidity needs or just to have a risk-free pile of assets to draw upon in times of financial stress. So where do investors find cash-like investments? For starters, investors can hold one of the safest shortterm cash-like instruments, which is a very short-maturity Treasury bill. This instrument is guaranteed by the U.S. government and at maturity it’s going to pay a fixed amount. Because of its very short maturity the price of a Treasury bill is hardly exposed to interest rate fluctuations. Therefore, it can be liquidated early with little interest rate risk. So Treasury bills satisfy investors’ need for safe overnight investments. However, the supply of Treasury bills is insufficient to meet the large demand for safe overnight investments. Therefore, cash pools must look for alternative cash-like investments. For example, investors can hold bank deposits. Bank deposits have some appealing characteristics. Like cash, the value of a bank deposit doesn’t fluctuate with interest rates and they can be used to meet payment obligations. However, bank deposits above a certain amount are uninsured, which exposes large investors to counterparty risk. If the bank fails, these investors are likely to lose money. Therefore, as opposed to safe Treasury bills, bank deposits are considered an unsecured overnight investment. Banks—also known as depository institutions—exchange funds overnight in the market for reserve balances at the federal funds rate, that is, the overnight unsecured interest rate that banks borrowing reserves have to pay to banks lending those reserves. Here it is worthwhile to further discuss the federal funds market and how it has changed since the 2008 financial crisis. The Federal Reserve (hereafter the Fed) interacts with the traditional banking sector by setting a target for the federal funds rate. For decades prior to the 2008 crisis, the target was achieved through temporary or permanent purchases or sales of government-guaranteed securities from and to banks. These transactions, 2 known as open market operations, aimed to adjust the supply of reserve balances in the banking system to create conditions that encourage the fed funds rate to trade at the target level. The Fed funds market consists of unsecured loans among banks and certain other eligible entities, 1 such as GSEs like the Federal Home Loan Banks, which have deposits at the Fed, but unlike private banks, GSEs do not get credited with any interest if they hold these balances overnight. Before the 2008 financial crises there were few excess reserves in the system and the Fed could change the equilibrium fed funds rate by conducting small open market operations, as this rate was very sensitive to small changes in the total level of excess reserves. During the financial crisis the Fed cut the federal funds target rate to a 0-basis-point through 25-basis-point range, and with its target rate at its effective zero lower bound, the Fed decided to expand the objective for open market operations to adjust the size and composition of its portfolio in order to stimulate the economy. These open market purchases of long-term government-guaranteed securities—known as Large Scale Asset Purchases—dramatically increased the amount of reserves held at the Fed by banks and other entities eligible to trade in the fed funds market. As a result, small changes in the amount of reserves outstanding no longer resulted in noticeable changes in the equilibrium federal funds rate. To better implement monetary policy in this new environment the Fed began paying interest on excess reserves (or IOER) to depository institutions in October 2008, but as previously mentioned, some institutions (such as the GSEs) are not eligible to earn IOER. Paying IOER influences the equilibrium federal funds rate by setting a floor on the rate at which depository institutions are willing to lend in the federal funds market. Banks that can earn a risk-free rate by holding reserves at the Fed should have no incentive to lend excess reserves in the fed funds market at rates below the IOER. But IOER has not served as a firm floor on the rate at which all institutions are willing to lend federal funds. One of the main reasons is that institutions such as GSEs do not earn IOER and therefore have an incentive to lend extra reserves at a rate below the IOER, because a small positive rate is better than nothing. As we mentioned earlier, there aren’t enough short-term Treasury bills to meet demand for cash-like riskfree investments. Therefore, cash pools are forced to look for alternative, private, cash-like safe investments—things that have the characteristics of cash but are not guaranteed by the full faith and credit of the United States government. Private agents produce these assets to meet the demand for cash-like investments. One of the more prominent examples of private money-like claims is shares issued by government-only money market mutual funds. Money market funds offer shares that have cash-like characteristics. The funds promise to redeem the shares for cash on demand at a fixed price. How does the money market fund do this? One way is the fund invests its pool of money in slightly longer-term Treasury securities. These Treasury securities have no default risk but are nonetheless riskier because their promised payments are further in the future, so changes in interest rates can change the present value (or market price) of these 1 See 204.2 (a)(1)(vii)(A) 3 bonds. Another option for cash pools looking for safe overnight cash-like investments is the overnight Treasury repurchase agreement (or the repo market). A repo is a sale of securities coupled with an agreement to repurchase the same securities at a higher price the next day. From the perspective of the cash pools investing money, a repo closely resembles a collateralized loan. For now, let’s focus on repos where Treasury securities are used as collateral. Broker dealers hold many longer-maturity Treasuries and often wish to use these bonds to obtain cheap funding in the repo market. Cash pools are willing to lend broker-dealers money at low rates because overnight repo loans are secured by Treasury collateral, and therefore very safe. However, repo lending does expose a cash pool to some counterparty risk. If the other side of the repo defaults, the cash pool will have to sell the securities to get its money back. However, the security is a safe government bond that can be sold easily. The lenders in a repo transaction further protect themselves by marking down the value of the collateral purchased relative to its true market value. This adjustment is known as a haircut. For example, the lender may agree to pay $100 today for securities that have $103 of market value. If the haircut is big enough, the lender will be able to resell the securities for more than the initial purchase price and suffer no loss. Similarly, money market funds wishing to lend money safely overnight can enter into repo agreements with broker-dealers. In practice, cash pools and money market funds lends money today by purchasing the government security and agreeing to sell it back to the original owner at a higher price tomorrow. The difference between the sale prices is the interest rate earned on the money lent overnight. The Repo market is an example of private agents creating an investment vehicle which transforms longerterm bonds into overnight debt instruments with safe cash-like characteristics. This is effectively private issuance of money-like assets that substitutes for public issuance of safe assets such as Treasury Bills. The repo market constitutes a very important part of what we call the “shadow banking market.” Before moving on, let’s take a moment to discuss an important institutional feature of the repo market—the tri-party system. For convenience reasons, many repo transactions take place through third party custodial banks, in what is known as the tri-party system. Custodial banks offer a series of services, including trade settlement, collateral valuation, and collateral eligibility screening. Many repo market participants prefer the tri-party system because the custodial banks create a platform that allows repo counterparties to come together and trade cheaply. The counterparties sign agreements ahead of time, specifying the terms of the repo transactions—such as what kind of collateral will be accepted and what haircuts will apply. The custodian then matches lenders to borrowers and holds the lent money and the securities pledged for the repo. For some time the Fed has used the tri-party system to conduct repurchase agreements with broker-dealers when conducting open market operations. Even after transforming government-guaranteed longer-term bonds into overnight instruments via repo, 4 there still may not be enough safe investments to meet the demand of cash pools. Some investors turn to other short-term securities that may not be as safe as overnight investments collateralized by public securities. One prominent example of a riskier short-term instrument is commercial paper. Commercial paper is a money-market security issued by corporations to obtain funds to meet short-term obligations. Commercial paper offers a higher return because it is backed by only the corporation’s promise to pay the face amount at maturity. Prime money market funds, like government only money market funds, promise overnight liquidity by redeeming their shares on demand at a fixed price. But, unlike government-only money market funds, prime funds invest in unsecured private assets that have some default risk. For example, they invest in commercial papers with high credit ratings and maturities of 90 days or less. Cash pools can invest directly in commercial paper or invest indirectly via prime money market funds, which offer better liquidity and easy access to a diversified portfolio of commercial paper. The combination of diversification and liquidity makes the indirect investments through prime money market funds an option with cash-like characteristics. Direct and indirect investment in commercial paper constitutes an important link between the shadow banking system and the real economy. Demand for commercial paper strongly affects the ability of corporations to raise funds to meet short-term debt obligations such as payrolls. In addition to private short-term securities, there are large amounts of private longer-term bonds and equities held by risk portfolio managers, hedge funds, real estate investment trusts and exchange-traded funds. These asset owners often wish to use their longer-maturity risky assets as collateral to obtain funding at a cheap rate. To this end, they enter into repo transactions with broker-dealers. To obtain short-term funding, these broker-dealers then re-pledge the acquired riskier collateral to prime money market funds that want money-like claims collateralized by private securities. Despite the riskier collateral, these private repos are close substitutes to money-like claims because they are overnight investments and have larger haircuts than Treasury repos. With a big enough haircut, securities such as government-agency mortgage-backed securities (MBS), private MBS, asset-backed securities, corporate bonds, or even equities can be transformed into assets that are viewed by market participants as safe overnight investments. Collectively, private repos make up a large portion of private money-like claims and are a prominent example of how money flows throughout the shadow banking system. As in the case of Treasury repos, many private repo transactions take place in the tri-party system where custodial banks offer the same services previously discussed. It is important to point out that even though the Fed has been operating in the tri-party system for some time, its actions have only indirectly influenced interest rates in the shadow banking system. 5 More specifically, by altering the fed funds rate, or more recently, the interest on excess reserves, the Fed changed the marginal cost of lending in the formal banking sector. Because most banks interact in both the formal and shadow banking sectors, raising the marginal cost of funds in one sector could affect the rate at which banks are willing to lend in the other sector. The extent to which a change in the federal funds rate is transmitted to interest rates in the shadow banking markets mainly depends upon banks’ perceived degree of substitution across different lending opportunities. You may have noticed that in what we have described so far, the Fed implemented monetary policy solely through interactions with bank counterparties. But, with the recent introduction of an overnight reverse repo facility, the Fed can now interact directly with a broader range of money market participants. These include both the Fed’s traditional counterparties, such as banks and broker dealers, and non-bank counterparties that play a major role in money markets. Importantly, this allows the Fed to provide a safe overnight investment to participants that cannot earn interest on excess reserves, effectively expanding the set of investors unwilling to lend at rates below those offered by the Fed. Let me explain in more detail. Due to Large Scale Asset Purchases, the Fed owns a large amount of safe government securities. In a reverse repo, the Fed sells government securities and borrows reserves from the eligible counterparties overnight at a fixed rate. This rate, known as the overnight reverse-repo rate is set by the Fed. Included among the non-bank counterparties eligible to participate in reverse repos with the Fed are the GSEs. Recall that the Fed funds rate often trades below IOER because institutions such as the GSEs are ineligible to earn IOER and are therefore willing to lend their excess reserves into the federal funds market at rates below IOER. By making GSEs eligible counterparties for overnight reverse repo transactions, the Fed is better equipped to influence rates in both the repo and the fed funds market. GSEs can now lend to the Fed at the fixed rate offered at the overnight reverse repo facility. Because these loans have the Fed as a counterparty and are collateralized by government securities, they are much safer than unsecured loans in the Fed funds market. Therefore, a GSE should be unwilling to lend in the fed funds market at a rate below the overnight reverse repo rate. Government-only and prime money market funds are also included among the non-bank counterparties eligible to participate in the overnight reverse repo facility. Here the same logic applies: These money market funds should be unwilling to lend to counterparties riskier than the Fed at a rate below that offered by the Fed at the overnight reverse repo facility. Overall, this implies that the rate set by the Fed at the overnight reverse repo facility should provide a floor on the level of short-term interest rates for cash-like investments in both the traditional and the shadow banking systems. Because, as we already discussed in the case of the GSEs and money market funds, active participants in overnight money markets should be unwilling to lend at a rate below the overnight reverse repo rate. 6 To sum up, these new tools should allow the Fed to better control short-term rates in both the traditional banking sector, by adjusting the interest on excess reserves, and the shadow banking sector, by adjusting the rate on reverse repos. As the FOMC indicated in the April 2015 Policy Normalization Principles and Plans, the Committee intends to target a 25-basis-point range for the federal funds rate. By altering the rates on IOER and overnight reverse repo, the Fed can adjust the upper and lower bounds of the targeted range. This should guide overnight money market rates to levels consistent with the Fed’s goals of maximum employment and stable prices. References Baklanova, Viktoria, Copeland, Adam, and McCaughrin, Rebecca (2015). “Reference Guide to U.S. repo and Securities Lending Markets,” Office of Financial Research Working Paper Series. Washington: Office of Financial Research, U.S. Department of the Treasury, http://financialresearch.gov/workingpapers/files/OFRwp-2015-17_Reference-Guide-to-U.S.-Repo-and-Securities-Lending-Markets.pdf. Bech, Morten, Elizabeth Klee, and Viktors Stebunovs (2014). "The Repo and Federal Funds Markets before, during, and Emerging from the Financial Crisis," in Chadha, Jagjit S., Alain C. J. Durre, Michael A. S. Joyce and Lucio Sarno eds., Developments in Macro-Finance Yield Curve Modelling. Cambridge, Mass: Cambridge University Press, pp. 293-325. Carlson, Mark, Burcu Duygan-Bump, Natalucci, Fabio, Nelson, William R., Ochoa, Marcelo, Stein, Jeremy, and Van den Heuvel, Skander (2014). "The Demand for Short-Term, Safe Assets and Financial Stability: Some Evidence and Implications for Central Bank Policies," Finance and Economics Discussion Series 2014-102. Washington: Board of Governors of the Federal Reserve System, http://www.federalreserve.gov/econresdata/feds/2014/files/2014102pap.pdf. Dudley, William C. (2014). “The Economic Outlook and Implications for Monetary Policy,” Remarks before the New York Association for Business Economics, New York City. New York: Federal Reserve Bank of New https://www.newyorkfed.org/newsevents/speeches/2014/dud140520. Dudley, William C. (2015). “The Economic Outlook and Implications for Monetary Policy,” Remarks before the New York Association for Business Economics, New York City. New York: Federal Reserve Bank of New York, https://www.newyorkfed.org/newsevents/speeches/2014/dud140520. Frost, Josh, Lorie Logan, Antoine Martin, Patrick McCabe, Fabio Natalucci, and Julie Remache (2015). “Overnight RRP Operations as a Monetary Policy Tool: Some Design Considerations,” Finance and Economics Discussion Series 2015-010. Washington: Board of Governors of the Federal Reserve System, http://dx.doi.org/10.17016/FEDS.2015.010. Ihrig, Jane E., Meade , Ellen E., and Weinbach, Gretchen C. (2015). “Monetary Policy 101: A Primer on the Fed’s Changing Approach to Policy Implementation,” Finance and Economics Discussion Series 2015-047. Washington: Board of Governors of the Federal Reserve System, http://dx.doi.org/10.17016/FEDS.2015.047. Potter, Simon (2014). “Implementation of Open Market Operations in a Time of Transition,” Remarks before the Japan Center for Economic Research, Tokyo, Japan. Tokyo: Federal Reserve Bank of New York, https://www.newyorkfed.org/newsevents/speeches/2014/pot140924. 7 Potter, Simon (2014). “Interest Rate Control during Normalization,” Remarks at the SIFMA Conference on Securities Financing Transactions, New York City. New York: Federal Reserve Bank of New York, https://www.newyorkfed.org/newsevents/speeches/2014/pot141007. Potter, Simon (2015). “Money Markets and Monetary Policy Normalization,” Remarks at the Money Marketeers of New York University, New York City. New York: Federal Reserve Bank of New York, https://www.newyorkfed.org/newsevents/speeches/2015/pot150415.html. Pozsar, Zoltan (2014). “Shadow Banking: The Money View,” Office of Financial Research Working Paper Series 14-04. Washington: Office of Financial Research, U.S. Department of the Treasury, http://financialresearch.gov/workingpapers/files/OFRwp201404_Pozsar_ShadowBankingTheMoneyView.pdf. The Chicago Fed Survey of Business Conditions: Quantifying the Seventh District’s Beige Book report Scott A. Brave, Thomas Walstrum, and Jacob Berman Introduction and summary The Beige Book is a Federal Reserve System report describing current business conditions in each of the Fed’s 12 Districts that is released to the public two weeks prior to each meeting of the Federal Open Market Committee (FOMC), the monetary policymaking arm of the Fed.1 Each Federal Reserve Bank prepares a report for its own District by surveying business contacts on topics such as demand for their products or services, capital spending, hiring, prices, and wages. Information collection methods for the Beige Book vary across the Reserve Banks, ranging from formal surveys to face-to-face interactions. By design, the Beige Book is an anecdotal, or qualitative, account that is meant to provide context for understanding trends in existing quantitative data. The timeliness of the report plays a key role in this function, as the gap between information collection and the public release of the report can be as short as one to two weeks. This feature of the Beige Book has led some researchers to try to quantify the information contained in the report so that it can be incorporated in real time into quantitative economic models. But because prose is open to different interpretations, it is far from obvious how to quantify the information. Thus, researchers have attempted a variety of techniques to quantify the language of the Beige Book, ranging from simple numerical scoring and word counts to more sophisticated analyses of linguistic patterns.2 In this article, we describe a new survey methodology used by the Federal Reserve Bank of Chicago in constructing its District’s Beige Book report called the Chicago Fed Survey of Business Conditions (CFSBC). The design of the survey allows us to create a new set of quantitative indexes that track economic activity in real time. The survey contains both quantitative and qualitative questions. We use answers to the quantitative questions to construct diffusion indexes that cover Federal Reserve Bank of Chicago a variety of aspects of economic activity, and we ask qualitative follow-up questions that provide context for the indexes and the Beige Book. The survey is timed to match the Beige Book schedule and has been operating since the Beige Book cycle for the March 6, 2013, report, although only a limited portion of the survey’s quantitative results have been published prior to this article.3 Survey respondents represent a wide range of industries in the Seventh Federal Reserve District,4 including construction, finance, manufacturing, real estate, retail, and a variety of other service industries. Scott A. Brave is a policy economist, Thomas Walstrum is a business economist, and Jacob Berman is a former associate economist in the Economic Research Department at the Federal Reserve Bank of Chicago. The authors thank Spencer Krane, Lisa Barrow, Daniel Sullivan, William Testa, William Strauss, and Gadi Barlevy for their helpful comments and suggestions. © 2015 Federal Reserve Bank of Chicago Economic Perspectives is published by the Economic Research Department of the Federal Reserve Bank of Chicago. The views expressed are the authors’ and do not necessarily reflect the views of the Federal Reserve Bank of Chicago or the Federal Reserve System. Charles L. Evans, President; Daniel G. Sullivan, Executive Vice President and Director of Research; David Marshall, Senior Vice President and Associate Director of Research; Spencer Krane, Senior Vice President and Senior Research Advisor; Daniel Aaronson, Vice President, microeconomic policy research; Jonas D. M. Fisher, Vice President, macroeconomic policy research; Robert Cox, Vice President, markets team; Anna L. Paulson, Vice President, finance team; William A. Testa, Vice President, regional programs; Lisa Barrow, Senior Economist and Economics Editor; Helen Koshy and Han Y. Choi, Editors; Julia Baker, Production Editor; Sheila A. Mangler, Editorial Assistant. Economic Perspectives articles may be reproduced in whole or in part, provided the articles are not reproduced or distributed for commercial gain and provided the source is appropriately credited. Prior written permission must be obtained for any other reproduction, distribution, republication, or creation of derivative works of Economic Perspectives articles. To request permission, please contact Helen Koshy, senior editor, at 312-322-5830 or email Helen.Koshy@chi.frb.org. ISSN 0164-0682 77 TABLE 1 CFSBC respondent composition percent Banking and finance 17 Construction and real estate 18 Manufacturing 32 Nonfinancial services 34 Note: The average number of respondents to the Chicago Fed Survey of Business Conditions (CFSBC) is 75. Source: Authors’ calculations. A number of organizations conduct similar surveys, such as the Institute for Supply Management (ISM) and other Federal Reserve Banks. However, the design of the Chicago Fed survey and the method used to construct the diffusion indexes differ from what is done for other business surveys in two important respects. First, the CFSBC asks respondents about both current and expected economic activity. And second, the resulting diffusion indexes are adjusted for inherent biases in measurement and interpretation. We do this by calculating the diffusion indexes based on whether survey participants’ answers are above or below their respective average answers. Expressing the indexes in this way is novel and is possible because most CFSBC respondents regularly participate in the survey. While the indexes remain works in progress, we believe that in their current form they offer useful insight into current conditions in the Seventh District and U.S. economies. Beginning with the release of the January 13, 2016, Beige Book, the Chicago Fed will make the CFSBC indexes publicly available at https://www.chicagofed.org/cfsbc. In what follows, we provide details on the CFSBC and the diffusion indexes we calculate from it. We first describe the survey’s quantitative questions and then explain how we turn them into diffusion indexes. Next, we explore how the unique adjustments we make to the formula for a traditional diffusion index affect its properties. We then assess how well our headline index, the CFSBC Activity Index, aligns with other indicators of economic activity. Finally, we discuss the additional information collected in the course of the survey that covers hiring, capital expenditures, and cost pressures. The survey and diffusion indexes Beginning with the cycle for the March 6, 2013, Beige Book, the Chicago Fed has been gathering information from its business contacts using an online survey system. The Bank’s regional analysis staff oversees the operation of the survey and invites individuals to join based on interactions at business roundtables, advisory councils, board meetings, conferences, speeches, and 78 other Bank events. Table 1 shows that survey respondents come from a variety of industries, with the largest representation coming from the nonfinancial services sector and the manufacturing sector. Over 600 invitations to participate in the survey are sent out each Beige Book reporting period, with a typical response rate of about 17 percent per survey. About 70 percent of survey participants are repeat responders. The survey asks respondents to quantify how aspects of their businesses have changed over the past four to six weeks (or are expected to change in the next six to 12 months) using a seven-point scale: increased substantially (+3), increased moderately (+2), increased slightly (+1), no change (0), decreased slightly (–1), decreased moderately (–2), and decreased substantially (–3). The survey covers a variety of topics, including product demand, prices, and productivity. We also ask sector-specific questions. For example, we ask our real estate contacts about changes in home prices. As a follow-up to each quantitative question, we ask respondents to provide anecdotes that explain why they answered the way they did. We then use this combination of quantitative and qualitative data as background material for the Seventh District’s Beige Book report. Using the responses to the quantitative questions in the survey, we construct diffusion indexes in order to track growth in economic activity in real time. The appendix lists the set of survey questions used to construct the indexes. The indexes cover the following topics: 1) overall activity (based on demand for products and services), 2) manufacturing activity, 3) nonmanufacturing activity, 4) the outlook for the U.S. economy, 5) hiring, 6) hiring plans, 7) capital spending, 8) capital spending plans, 9) wage costs, and 10) nonwage costs. Diffusion indexes are intended to be leading indicators, capturing changes in the prevailing direction of economic activity. The formula for CFSBC diffusion indexes is #Above-Average (Positive) Responses – #Below-Average (Negative) Responses . 100 * #Responses (smoothed) In many ways, this is a very traditional formula for a diffusion index, but we make a couple of adjustments that we would argue improve it. The first and most novel adjustment is to measure individuals’ responses relative to their respective average responses.5 To calculate a respondent’s average response, we assign numerical values ranging from +3 to –3 along a seven-point scale, and take the average across all responses, including the current one. We then count a response as positive if it is above a respondent’s average response and negative if it is below a respondent’s average response. 3Q/2015, Economic Perspectives FIGURE 1 CFSBC Activity Index index 80 60 40 20 0 –20 –40 –60 2013 ’14 ’15 As published (detrended) Not detrended Notes: CFSBC means Chicago Fed Survey of Business Conditions. The detrended version will be published in the official Chicago Fed release of the survey’s results. See the text for details on the detrending methodology. Source: Authors’ calculations. For example, if a respondent’s average response is +1.5, substantial and moderate increases are counted as positive responses and all other answers are counted as negative responses. Given our formula, the index ranges from +100 to –100 and will be +100 if every respondent in a given survey has an above-average response to a question and –100 if every respondent has a below-average response.6 We do not include respondents’ answers in the index until they have answered the survey twice. This is because those who respond only once would have neutral answers by definition, as a single response is necessarily the average response. Note, too, that over time, the index could change because respondents’ averages will change with each response. Of course, the longer the history for a respondent, the more stable his average will be, so that this source of variation will become less important over time. And, as discussed later, in practice, changes in average responses do not appear to be an important concern in interpreting the CFSBC indexes. Calculating the indexes using a survey participant’s average response as a baseline—also known as detrending—allows us to correct for two types of potential biases. First, individuals may interpret phrases such as “substantially increased” differently, so that our numerical scores have different meanings for different people. A common criticism of the Beige Book is that the wording used to describe changes in economic activity seldom varies and can often be so broad as to make it difficult to interpret differences across regions and industries. So we rely upon individuals’ own assessments of “average” growth in order to assess how far from “normal” Federal Reserve Bank of Chicago business conditions currently are. This process amounts to a rescaling of the responses to the survey questions that is unique to each respondent prior to constructing what would otherwise be a traditional diffusion index. The second type of potential bias is that the industries and firms represented in our data may have different growth trends than the overall economy, which could bias the indexes because we do not have a random sample of respondents. For example, because manufacturers represent a significant share of our respondents, our index could overrepresent trends in the manufacturing sector. The share of manufacturing output and employment in the U.S. economy has been declining for decades, so that in general, trends in the manufacturing sector are becoming less and less representative of trends in the overall economy. For this reason, other similar surveys often report separate manufacturing and nonmanufacturing indexes and either refrain from combining them into a single index or weight them based on their representativeness. However, because calculating a CFSBC index relative to respondents’ average responses removes respondent-, firm-, and industry-specific trends, we can safely combine manufacturers and nonmanufacturers into the same respondent pool and mitigate some of the bias from having a nonrandom sample of respondents. Like the researchers conducting similar surveys, we calculate manufacturing and nonmanufacturing indexes separately, and these indexes also benefit from detrending because it removes respondent- and firm-specific trends. Another adjustment we make to the standard diffusion index formula is to smooth the denominator, which is the total number of repeat responses in a given period. We do this by calculating the denominator as a threeperiod moving average of the lag period, the current period, and the lead period.7 We do this because there are a few periods over the history of the index where there was a notable increase in the number of repeat respondents.8 A large increase in repeat respondents from one period to the next could cause a change in the value of an index that is unrelated to a change in the distribution of responses. To better understand why, consider the following scenario. Suppose that there are two adjacent periods where, initially, all respondents are present in both periods and their responses are identical in both periods. In this case, the diffusion index will indicate that activity did not change from the first period to the second. Now suppose that we add 20 new repeat respondents in the second period and that ten give above-average reports and ten give below-average reports. In this scenario, the numerator of the index is unaffected, but the denominator is larger, so that the diffusion index is now lower in the second period, even 79 FIGURE 2 Distribution of responses, by type, for the CFSBC Activity Index A. Detrended B. Not detrended frequency frequency 800 800 600 600 400 400 200 200 0 negative neutral positive 0 negative neutral positive Notes: CFSBC means Chicago Fed Survey of Business Conditions. Panel A shows the distribution of responses, by type, for the detrended index—which is the CFSBC Activity Index that will be published in the official Chicago Fed release of the survey’s results (also shown as the blue line in figure 1). See the text for details on the detrending methodology. Source: Authors’ calculations. though the difference between above-average and belowaverage responses is unchanged from the first period. We would view such a decline in the index as misleading. Smoothing the denominator helps to mitigate the distortion, though not perfectly. In practice, we find little difference between our preferred versions of the indexes (which will be published in the official Chicago Fed release of the survey’s results) and versions without a smoothed denominator. How does detrending influence the CFSBC diffusion indexes? To explain the implications of our detrending methodology, we examine how two of our indexes, the CFSBC Activity Index and CFSBC Outlook Index, would differ if we instead calculated them using the traditional diffusion index formula. Figure 1 (p. 79) shows the CFSBC Activity Index and a version calculated from the traditional formula. The activity index is based on respondents’ answers to questions about demand for their products or services. Respondents typically report that demand increased slightly, so detrending shifts the index down. In general, the shapes of the paths of both indexes are quite similar: Both indexes capture the slowdowns in the first quarters of 2014 and 2015 and the strength of growth in the 80 second half of 2014. That said, compared with the version calculated from the traditional formula, the official CFSBC Activity Index (which has been detrended) indicates that the downturns were sharper. This is not surprising, since detrending will increase the variance of respondents’ answers if their answers always fall in a limited range (in this case mostly positive). To clarify why the variance of the index increases when the index is detrended, we show in figure 2 the distribution of positive, neutral, and negative responses to the demand questions across all survey rounds before and after detrending. Because the average response for many respondents is that demand increased slightly, detrending converts many of the neutral responses into negative responses. There are also some respondents who typically report a moderate increase in demand, so that detrending converts responses indicating a slight increase in demand into negative responses. The new distribution created by detrending is more spread out, increasing its variance, and, not surprisingly, increasing the variance of the diffusion index as well. As noted in the previous section, detrending is also useful because it adjusts for differences in the respondents’ interpretations of the quantitative scale that we use. For example, some respondents may be more optimistic or pessimistic in general, and detrending 3Q/2015, Economic Perspectives FIGURE 3 FIGURE 4 CFSBC Outlook Index Respondents eligible for the CFSBC Activity Index, by minimum response requirement index number of respondents 80 100 60 80 40 60 20 40 0 20 –20 2013 ’14 0 2013 ’15 ’14 As published (detrended) Not detrended Notes: CFSBC means Chicago Fed Survey of Business Conditions. The detrended version will be published in the official Chicago Fed release of the survey’s results. See the text for details on the detrending methodology. Source: Authors’ calculations. (or, more appropriately, demeaning) allows us to put them on equal footing. This particularly applies to when the survey asks for respondents’ outlooks for the U.S. economy over the next six to 12 months. Detrending makes it so that our moderately optimistic respondents are counted as having a positive outlook only when they are feeling very optimistic and our moderately pessimistic respondents are counted as having a positive outlook even when they are feeling slightly pessimistic. Figure 3 shows the CFSBC Outlook Index and a version constructed in the traditional way (that is, without detrending). Here, as with the CFSBC Activity Index, respondents typically report that they have a positive outlook, so the line for the official CFSBC Outlook Index is below the line for the index without detrending. Moreover, the shapes of the paths of both indexes are quite similar, though the detrended index has a greater variance than the index constructed using the traditional formula. In other words, we see that the two outlook indexes in figure 3 behave similarly to the two activity indexes in figure 1 (p. 79). As noted previously, detrending also leads us to exclude survey participants who have responded only once to the survey from the respondent pool because all of their responses are neutral by definition. In addition, we expect that survey participants’ average responses will change as their response histories grow. The longer the history is, the more stable the average response is—and the more reliably we can identify deviations from a long-run trend. Thus, because some survey participants respond more consistently than others do, we face a trade-off between maximizing Federal Reserve Bank of Chicago ’15 No minimum At least two responses At least three responses At least four responses Note: CFSBC means Chicago Fed Survey of Business Conditions. Source: Authors’ calculations. FIGURE 5 Distribution of the number of times responded, by survey respondent, for the CFSBC Activity Index percent 25 20 15 10 5 0 0 5 10 15 20 number of times responded Notes: CFSBC means Chicago Fed Survey of Business Conditions. As of November 2015, there are 264 unique respondents and 23 unique survey periods. Source: Authors’ calculations. the quantity of respondents and the quality of responses in calculating the indexes. Figure 4 shows how the number of respondents used to calculate the CFSBC Activity Index declines as we increase the minimum number of responses per respondent required. The vast majority of survey participants who contributed to the index prior to 2015 have responded at least four times. Throughout 2015, we have added a number of new survey participants, and most of them have not yet responded four times. Figure 5 provides another perspective on the CFSBC Activity Index’s respondent pool. It shows the distribution of the number of survey responses by respondent as of 81 FIGURE 6 FIGURE 8 CFSBC Activity Index, by minimum response requirement CFSBC Activity Index and the MEI index 60 40 index index 50 0.87 25 0.57 0 0.27 20 0 –20 –25 –0.03 –40 –60 2013 ’14 ’15 At least two responses At least three responses At least four responses Note: CFSBC means Chicago Fed Survey of Business Conditions. Source: Authors’ calculations. FIGURE 7 Revisions to the CFSBC Activity Index index 60 40 20 0 –20 –40 –60 2013 ’14 ’15 Range of previous values Current values Note: CFSBC means Chicago Fed Survey of Business Conditions. Source: Authors’ calculations. November 2015. Just under 75 percent of the respondent pool (made up of 264 unique respondents) has responded at least two times, allowing that share to be included in the least restrictive version of the index, which is our preferred version (displayed as the blue line in figure 1 on p. 79). Figure 6 shows how the CFSBC Activity Index would change if we altered the minimum number of responses required for a survey participant’s response to be counted in the calculation of the index. Increasing the requirement from two to four has little effect on the index except at the very beginning when the respondent pool was small in comparison to its current size. There are some differences in 2015 related to the large number of new respondents without a long history, 82 –50 2013 ’14 ’15 –0.33 CFSBC Activity Index (left-hand scale) MEI (right-hand scale) Notes: CFSBC means Chicago Fed Survey of Business Conditions. MEI means the Midwest Economy Index. We align the MEI’s average over the period March 2013– September 2015 with a value of zero from the CFSBC Activity Index to facilitate the comparison. The average of the MEI over the reference period is 0.27. The correlation coefficient between the indexes is 0.51. Source: Authors’ calculations. but they are small and do not substantially change how we would interpret the index. A bigger concern is the fact that as we add to the histories of our respondents, their average responses evolve, sometimes changing whether we treat earlier responses as positive, neutral, or negative, which in turn changes earlier values of the CFSBC indexes. Figure 7 shows the current CFSBC Activity Index as well as the range of values it has had in the past. The range of values provides some indication of how much more-recent values may change as future survey rounds are completed. In our view, the ranges are small enough that they have little effect on how one would interpret the index’s value. Moreover, we expect the size of such revisions to decline in the coming years as we continue to add to our respondents’ histories and their respective average responses become more stable.9 Comparing the CFSBC Activity Index with other measures of activity In this section, we explore how well the CFSBC Activity Index aligns with other indicators of economic activity at the District and national level, such as the Chicago Fed’s Midwest Economy Index (MEI), the Chicago Fed National Activity Index (CFNAI), and the national surveys of manufacturing and nonmanufacturing activity conducted by the Institute for Supply Management. The CFSBC Activity Index (as well as its manufacturing and nonmanufacturing subindexes) largely traces a path similar to the paths of these indicators, but 3Q/2015, Economic Perspectives FIGURE 9 FIGURE 10 CFSBC Activity Index and the CFNAI CFSBC Manufacturing Activity Index and the ISM manufacturing PMI index index 50 0.52 25 0.28 0 0.04 index index 50 60 25 57 0 54 –25 51 –0.20 –25 –50 2013 –0.44 ’14 ’15 CFSBC Activity Index (left-hand scale) CFNAI (right-hand scale) Notes: CFSBC means Chicago Fed Survey of Business Conditions. CFNAI means the Chicago Fed National Activity Index. We align the CFNAI’s average over the period March 2013– September 2015 with a value of zero from the CFSBC Activity Index to facilitate the comparison. The average of the CFNAI over the reference period is 0.04. The correlation coefficient between the indexes is 0.49. Source: Authors’ calculations. the timeliness of the information it provides gives it an advantage over some of these indicators. We also show that the CFSBC Activity Index correlates quite well with the Seventh District’s real gross state product (GSP) growth and U.S. real gross domestic product (GDP) growth, suggesting that the index may have some value to researchers interested in forecasting these measures in real time. Figure 8 presents the CFSBC Activity Index and the Midwest Economy Index together. The MEI is a weighted average of 129 Seventh District state and regional indicators measuring growth in nonfarm business activity from four broad sectors of the Midwest economy: manufacturing, construction and mining, services, and consumer spending.10 Thus, it summarizes a wide range of information about economic activity in the Seventh District. Because the MEI has been above its long-run trend for most of the time since March 2013, we equate the MEI’s average over this reference period11 to a value of zero from the CFSBC Activity Index to facilitate the comparison. The CFSBC Activity Index and MEI demonstrate considerable co-movement, such as the ramp up in activity in the middle of 2014 and the more recent slowdown. However, there are also times when they move in different directions. One possible explanation for this is that the MEI tends to be subject to large revisions, as much of its underlying data series are subject to benchmark revisions. In previous work, we showed that an early version of the CFSBC Activity Index was Federal Reserve Bank of Chicago –50 2013 48 ’14 ’15 CFSBC Manufacturing Activity Index (left-hand scale) ISM manufacturing PMI (right-hand scale) Notes: CFSBC means Chicago Fed Survey of Business Conditions. ISM means the Institute for Supply Management; PMI means purchasing managers’ index. We align the ISM manufacturing PMI’s average over the period March 2013–October 2015 with a value of zero from the CFSBC Manufacturing Activity Index to facilitate the comparison. The average of the ISM manufacturing PMI over the reference period is 54. The correlation coefficient between the indexes is 0.52. Sources: Authors’ calculations and Institute for Supply Management data from Haver Analytics. predictive of the direction of these revisions in recent years.12 Another possible explanation is that the CFSBC Activity Index is capturing a combination of regional and national economic activity because many of the firms in the survey sample have a national footprint. Figure 9, therefore, compares the plots of the CFSBC Activity Index and the Chicago Fed National Activity Index, which is a weighted average of 85 national indicators measuring growth in production and income; employment, unemployment, and hours; personal consumption and housing; and sales, orders, and inventories.13 As with the MEI comparison, we equate the CFNAI’s average over the reference period to a value of zero from the CFSBC Activity Index to facilitate the comparison. Comparing figures 8 and 9, one can see that at least some of the periods of divergence between the CFSBC Activity Index and the MEI align with periods where the CFSBC Activity Index instead more closely follows the CFNAI. Overall, the CFSBC Activity Index appears to capture similar information on growth in economic activity as the MEI and CFNAI, which is confirmed by its correlation coefficients of 0.51 with the MEI and 0.49 with the CFNAI. It is important to keep in mind, though, that the CFSBC Activity Index is timelier than the MEI and CFNAI. New releases of the CFSBC indexes will typically come out one to two weeks after the survey is completed, while both the MEI and CFNAI are released with a 83 FIGURE 11 FIGURE 12 CFSBC Nonmanufacturing Activity Index and the ISM nonmanufacturing PMI CFSBC Activity Index and real GSP growth for the Seventh Federal Reserve District index index 60 61.0 30 58.5 0 56.0 –30 53.5 –60 2013 51.0 ’14 ’15 CFSBC Nonmanufacturing Activity Index (left-hand scale) ISM nonmanufacturing PMI (right-hand scale) Notes: CFSBC means Chicago Fed Survey of Business Conditions. ISM means the Institute for Supply Management; PMI means purchasing managers’ index. We align the ISM nonmanufacturing PMI’s average over the period March 2013–October 2015 with a value of zero from the CFSBC Nonmanufacturing Activity Index to facilitate the comparison. The average of the ISM nonmanufacturing PMI over the reference period is 56. The correlation coefficient between the indexes is 0.58. Sources: Authors’ calculations and Institute for Supply Management data from Haver Analytics. one-month lag. Thus, while the CFSBC Activity Index contains similar information as the MEI and CFNAI, it can signal movements in Seventh District and national economic activity sooner. We next compare the CFSBC Manufacturing and Nonmanufacturing Activity Indexes with the Institute for Supply Management’s manufacturing and nonmanufacturing purchasing managers’ indexes (PMIs).14 The ISM’s PMIs are quite similar to the CFSBC indexes in that they are also diffusion indexes constructed from a survey that asks purchasing and supply executives whether their production, employment, orders, and inventories are higher or lower than (or the same as) in the previous reporting period. While ISM PMIs exist for some cities and regions in the Seventh District, the ISM only calculates manufacturing and nonmanufacturing indexes separately at the national level. Figure 10 (p. 83) shows the CFSBC Manufacturing Activity Index and ISM manufacturing PMI together. In this figure, we equate the average value of the ISM manufacturing PMI over the reference period to the baseline of the CFSBC Manufacturing Activity Index to facilitate the comparison. The indexes largely move together, with the exception of the first half of 2013. In addition, the ISM manufacturing PMI declined faster than the CFSBC Manufacturing Activity Index did in the first half of 2015, likely because of strong 84 index percent 60 8.8 40 6.6 20 4.4 0 2.2 –20 0.0 –40 –2.2 –60 2013 –4.4 ’14 ’15 CFSBC Activity Index (left-hand scale) Real GSP growth (right-hand scale) Notes: CFSBC means Chicago Fed Survey of Business Conditions. GSP means gross state product. Seventh District GSP is the sum of the GSP of the Seventh District states (Illinois, Indiana, Iowa, Michigan, and Wisconsin). We align the Seventh District’s average real GSP growth over the period 2013:Q1–2014:Q4 with a value of zero from the CFSBC Activity Index to facilitate the comparison. The average quarterly (annualized) real GSP growth rate over the reference period is 2.2 percent. The correlation coefficient between the index and the growth rate is 0.54. Sources: Authors’ calculations and U.S. Bureau of Economic Analysis data from Haver Analytics. growth in the Seventh District’s auto industry. Figure 11 shows the CFSBC Nonmanufacturing Activity Index and the ISM nonmanufacturing PMI together; similar to previous comparisons, we equate the average value of the ISM nonmanufacturing PMI over the reference period to the baseline of the CFSBC Nonmanufacturing Index. They, too, generally move together. Similar to what we saw for the CFSBC Activity Index vis-à-vis the MEI and the CFNAI, the co-movement of the CFSBC activity indexes and the ISM PMIs is notable: The correlation coefficient between the manufacturing indexes is 0.52, and the correlation coefficient between the nonmanufacturing indexes is 0.58. Figures 10 and 11 suggest that the CFSBC Activity Index also aligns quite well with other survey measures of economic activity. This finding highlights one advantage that the overall CFSBC Activity Index has over the individual ISM indexes: It efficiently combines the information from both manufacturers and nonmanufacturers into a single index that accounts for the differences in trend growth in each sector.15 Another advantage of the CFSBC Activity Index over the ISM’s PMIs is that it shares much conceptually with traditional measures of gross output because of its focus on the demand for firms’ products and services. Figures 12 and 13 explore this relationship by comparing the CFSBC Activity Index with the real 3Q/2015, Economic Perspectives FIGURE 13 CFSBC Activity Index and U.S. real GDP growth index percent 70 7.2 35 4.8 0 2.4 –35 0.0 –70 2013 –2.4 ’14 ’15 CFSBC Activity Index (left-hand scale) Real GDP growth (right-hand scale) Notes: CFSBC means Chicago Fed Survey of Business Conditions. GDP means gross domestic product. We align average real GDP growth over the period 2013:Q1–2015:Q3 with a value of zero from the CFSBC Activity Index to facilitate the comparison. The average quarterly (annualized) real GDP growth rate over the reference period is 2.4 percent. The correlation coefficient between the index and the growth rate is 0.54. Sources: Authors’ calculations and U.S. Bureau of Economic Analysis data from Haver Analytics. growth rates of Seventh District GSP16 and U.S. GDP. In these figures, we have aligned the average of the gross output measures over the reference period with the baseline of the CFSBC Activity Index. One can see from the figures that the CFSBC Activity Index and growth in both gross output measures often move together, with periods where the activity index is above zero tending to correspond with quarters characterized by above-average gross output growth and vice versa. While the time series of the CFSBC Activity Index is not quite long enough to allow for a formal test of these relationships, the strong correlations in these figures do suggest that the index might be a valuable input into nowcasting models of both gross output measures.17 Additional current and forward-looking indexes We also use the CFSBC to construct current and forward-looking diffusion indexes for a number of additional topics typically covered in the Beige Book: current and expected hiring and capital spending, as well as wage and nonwage cost pressures. These indexes are also constructed using detrended responses, so that they share the same interpretation as that for the activity and outlook indexes—where a positive value reflects aboveaverage growth and a negative value reflects belowaverage growth. Figure 14 presents all of these indexes, and also includes the latest results from follow-up questions that we ask about the types of workers firms are looking to hire, the types of capital spending firms Federal Reserve Bank of Chicago are undertaking, and the types of wage and nonwage cost pressures firms are facing. These additional indexes, particularly those related to hiring and capital spending, provide further context for the responses underlying the CFSBC Activity and Outlook Indexes shown in the top left panel of figure 14. Both the indexes for current hiring and capital spending tend to rise and fall with the CFSBC Activity Index, while the indexes for hiring and capital spending plans (for the next six to 12 months) tend to more closely follow the CFSBC Outlook Index. The current hiring and capital spending indexes are also highly reflective of the mild nature of the expansion in activity over the past few years, with both measures generally exhibiting below-average growth for most of this time. Similarly, the muted inflation that has persisted during this period is consistent with the below-average index values for both wage and nonwage cost pressures. The bar charts in figure 14 provide further information about the forces driving the values of the indexes. When tracked over time, they, too, provide valuable information for the Seventh District’s Beige Book report. For instance, when asked to identify the types of occupations that their firms are hiring for, respondents have long indicated stronger demand for professional and technical occupations than for other categories. Consistent with this, respondents tend to report greater wage pressure for such occupations. Respondents have also tended to describe capital spending as being confined mostly to replacement of existing information technology and industrial equipment. As both hiring and capital spending trended up in 2014, respondents noted a broadening in the types of occupations in demand, sources of wage pressures, and types of capital spending. More-recent values of these indexes reflect a weakening of business conditions in 2015. Conclusion The Chicago Fed Survey of Business Conditions offers a new set of diffusion indexes for tracking economic activity in real time. The design of the survey and the method used to construct the diffusion indexes are unique in that the survey asks respondents about both current and expected economic activity and the resulting diffusion indexes are adjusted for inherent biases in measurement and interpretation. By calculating our diffusion indexes based on whether participants’ answers are above or below their respective average answers, we address potential biases from respondent-, firm-, and industry-specific trends that arise because our respondent pool is not a random sample. The correlation of the resulting diffusion indexes with other regional and national indexes of economic activity, as well as 85 FIGURE 14 CFSBC diffusion indexes and related data Activity Hiring index 60 Occupations (percent) index 60 Current Outlook Current Plans Management 40 40 Professional & technical 20 20 Sales 0 44 35 Administration 0 0 30 19 Maintenance –20 –16 –20 –40 –60 2013 –21 –33 –40 ’14 ’15 12 Production Transportation 19 4 Other –60 2013 ’14 16 ’15 Capital spending Spending for (percent) index 60 Current Plans Spending on (percent) Replacement 40 Expansion 20 Research & development Mergers & acquisitions 0 65 Industrial equipment 34 IT equipment 41 49 Transportation equipment 29 8 Structures 6 35 Intellectual property –20 –25 –26 28 Other 12 –40 –60 2013 ’14 ’15 Cost pressures Increasing wage costs (percent) index 60 40 Wage Nonwage Management Sales 0 Administration –20 –18 Maintenance –19 58 16 Transportation Other 2013 ’14 Energy 6 Equipment 6 12 Benefits 28 35 Shipping 7 Taxes & regulations 7 Other –60 16 Property 23 Production –40 Materials 47 Professional & technical 20 Increasing nonwage costs (percent) 69 4 35 6 ’15 Notes: CFSBC means Chicago Fed Survey of Business Conditions. This multipaneled figure will be featured on the back page of the official Chicago Fed release of the survey’s results. For details on how to interpret the various line graphs, see the text. The bar graphs report the percentage of responses for types of occupations currently in demand; reasons for capital spending and its allocation; and sources of wage and nonwage cost pressures. The numbers may add up to more than 100 percent because more than one option can be chosen. Source: Authors’ calculations. 86 3Q/2015, Economic Perspectives gross output growth, is solid. Additionally, the CFSBC diffusion indexes offer insight into the hiring and capital spending decisions of business contacts surveyed for the Seventh District’s Beige Book report. As we have demonstrated, the indexes in their current form offer useful insights into current conditions in the Seventh District and U.S. economies. However, they remain works in progress. For example, as the time range of observation grows, we will be able to adjust for potential seasonality in the indexes. We also plan to calculate standard errors for the indexes based on work by Pinto, Sarte, and Sharp (2015). Finally, we may be able to release indexes based on other questions we ask our respondents, such as those regarding prices, productivity, and credit conditions. NOTES More specifically, the Beige Book is always released on a Wednesday two weeks ahead of the final day of an FOMC meeting. For the official description of the Beige Book and to view the latest report as well as the archives, visit https://www.federalreserve.gov/ monetarypolicy/beigebook/. Information about the 12 Federal Reserve Districts and Banks is available at https://www.federalreserve.gov/ otherfrb.htm. 1 Balke and Petersen (2002) go through 14 years of Beige Book reports and assign numerical scores (ranging from –2 to 2) to the level of growth described in the text. Both authors read passages in a random order and use the average of their two scores. They then use their measure to predict current and next quarter real gross domestic product (GDP) growth and are able to outperform the consensus forecasts in the Blue Chip Economic Indicators for some specifications. Armesto et al. (2009) use linguistics software to measure the degree to which language in the Beige Book is either optimistic or pessimistic. They employ the resulting index in an econometric model that explicitly accounts for the irregular schedule of the Beige Book. They find that the national Beige Book summary can predict both GDP and employment growth. They also find that the District Beige Book reports can predict regional employment growth. Balke, Fulmer, and Zhang (2015) use textual analysis techniques developed in Fulmer (2014) to quantify Beige Book language. They find that their measure provides unique information about current economic conditions, particularly during recessions, in the context of a dynamic factor model of U.S. business conditions. 2 In a related article, Brave and Walstrum (2014) introduce a subset of these indexes and show how they are useful for understanding turning points in economic activity. 3 The Seventh Federal Reserve District (which is served by the Chicago Fed) comprises all of Iowa and most of Illinois, Indiana, Michigan, and Wisconsin; for more details, see note 1 and https://www.chicagofed.org/utilities/about-us/seventh-district-economy. 4 This adjustment is akin to controlling for respondent fixed effects in terms of a traditional linear regression model. 5 While not a concern in practice, it is possible for the diffusion indexes to breach these bounds (of +100 and –100) given the smoothing procedure that we employ for the denominator in the formula. We explain the smoothing procedure a little later in this section (describing the survey and the construction of the diffusion indexes). 6 The denominator for the latest period is a two-period moving average of the current and lag period, since the lead period is unknown. 7 Federal Reserve Bank of Chicago Notable jumps in the number of repeat survey respondents (visible in figure 4 on p. 81) coincide with initiatives undertaken to increase the sample size of the survey. 8 It is worth noting, however, that the timing of respondents’ entrances into the survey relative to the business cycle will have an important effect on our estimate of their long-run trends. For example, we will underestimate the long-run growth trends of respondents who enter during a recession until we have sufficient observations of their responses during an expansion. If we add a large number of respondents to the survey when the economy is not growing near its long-term trend, this could bias the indexes. Because data for the indexes cover a period of growth that is near trend, we do not believe this concern is currently a major source of bias. 9 For more information on the MEI, see https://www.chicagofed.org/mei. 10 Reference periods will vary depending on the availability and frequency of the index or gross output measure to which a CFSBC index is compared in figures 8–13 (the precise periods appear in the figures’ notes). 11 12 See Brave and Walstrum (2014). For more information on the CFNAI, see https://www.chicagofed.org/ cfnai. 13 For more information on the ISM’s PMIs, see https://www. instituteforsupplymanagement.org/news/content.cfm?ItemNumber =28965&&navItemNumber=28882. 14 There are, however, other ways in which to combine the information from manufacturers and nonmanufacturers. For instance, some analysts have combined the ISM manufacturing and nonmanufacturing PMIs into a composite index by weighting them by the relative contribution of manufacturing and nonmanufacturing industries to GDP. 15 Seventh District GSP is the sum of the GSP of the Seventh District states (Illinois, Indiana, Iowa, Michigan, and Wisconsin). 16 The term nowcasting is derived from combining the words now and forecasting. Nowcasting techniques are commonly used in economics nowadays because they permit economists to predict today the present (and recent past) of standard measures of the economy (such as real GDP), which are often determined after a long delay. For more information on nowcasting real GDP growth with indexes of economic activity, see Brave and Butters (2014). For more information on forecasting real GSP growth, see Brave and Wang (2011). 17 87 APPENDIX: CFSBC DIFFUSION INDEX QUESTIONS 1. In the past four to six weeks, demand for my firm’s products or services has ○○ increased substantially. ○○ increased moderately. ○○ increased slightly. ○○ not changed. ○○ decreased slightly. ○○ decreased moderately. ○○ decreased substantially. 2. My outlook for the U.S. economy in the next six to 12 months is ○○ very positive. ○○ moderately positive. ○○ slightly positive. ○○ neutral. ○○ slightly negative. ○○ moderately negative. ○○ very negative. 3. In the past four to six weeks, my firm’s work force has ○○ increased substantially. ○○ increased moderately. ○○ increased slightly. ○○ not changed. ○○ decreased slightly. ○○ decreased moderately. ○○ decreased substantially. 4. In the next six to 12 months, I expect my firm’s work force to ○○ increase substantially. ○○ increase moderately. ○○ increase slightly. ○○ not change. ○○ decrease slightly. ○○ decrease moderately. ○○ decrease substantially. 88 5. My firm is hiring or looking to hire for these occupations: □□ managerial (executive, accountant, HR manager, marketing, etc.) □□ professional and technical (engineer, IT support, lawyer, etc.) □□ sales □□ administrative support □□ maintenance (mechanic, custodian, etc.) □□ production (operator, assembler, quality assurance, laborer, etc.) □□ transportation (driver, material handling, etc.) □□ other 6. In the past four to six weeks, my firm’s capital spending has ○○ increased substantially. ○○ increased moderately. ○○ increased slightly. ○○ not changed. ○○ decreased slightly. ○○ decreased moderately. ○○ decreased substantially. A. My firm’s capital spending has been for: □□ replacing equipment or remodeling structures □□ capacity expansion □□ research and development □□ mergers and acquisitions B. My firm’s capital spending has been on: □□ industrial equipment □□ IT equipment □□ transportation equipment □□ structures □□ intellectual property □□ other 7. In the next six to 12 months, I expect my firm’s capital spending to ○○ increase substantially. ○○ increase moderately. ○○ increase slightly. ○○ not change. ○○ decrease slightly. ○○ decrease moderately. ○○ decrease substantially. 3Q/2015, Economic Perspectives 8. In the past four to six weeks, my firm’s overall wage costs have ○○ increased substantially. ○○ increased moderately. ○○ increased slightly. ○○ not changed. ○○ decreased slightly. ○○ decreased moderately. ○○ decreased substantially. A. Please select the occupations for which wage costs have increased: □□ managerial (executive, accountant, HR manager, marketing, etc.) □□ professional and technical (engineer, IT support, lawyer, etc.) □□ sales □□ administrative support □□ maintenance (mechanic, custodian, etc.) □□ production (operator, assembler, quality assurance, laborer, etc.) □□ transportation (driver, material handling, etc.) □□ other 9. In the past four to six weeks, my firm’s overall nonwage costs have ○○ increased substantially. ○○ increased moderately. ○○ increased slightly. ○○ not changed. ○○ decreased slightly. ○○ decreased moderately. ○○ decreased substantially. A. Please select the areas where nonwage costs have increased: □□ raw materials or wholesale goods □□ energy □□ equipment □□ property □□ benefits (health insurance, retirement, etc.) □□ shipping costs □□ taxes and regulations □□ other REFERENCES Armesto, Michelle T., Rubén Hernández-Murillo, Michael T. Owyang, and Jeremy Piger, 2009, “Measuring the information content of the Beige Book: A mixed data sampling approach,” Journal of Money, Credit and Banking, Vol. 41, No. 1, February, pp. 35–55. Brave, Scott A., and Thomas Walstrum, 2014, “Using the Federal Reserve’s Beige Book to track economic activity,” Chicago Fed Letter, Federal Reserve Bank of Chicago, No. 328, November, available at https://www.chicagofed.org/publications/chicago-fedletter/2014/november-328. Balke, Nathan S., Michael Fulmer, and Ren Zhang, 2015, “Incorporating the Beige Book into a quantitative index of economic activity,” Southern Methodist University and Federal Reserve Bank of Dallas, mimeo, August 26, available at http://www.ssc.upenn.edu/ ~fdiebold/NoHesitations/Balkeetal2015.pdf. Brave, Scott A., and Norman Wang, 2011, “Predicting gross state product growth with the Chicago Fed’s Midwest Economy Index,” Chicago Fed Letter, Federal Reserve Bank of Chicago, No. 293, December, available at https://www.chicagofed.org/publications/ chicago-fed-letter/2011/december-293. Balke, Nathan S., and D’Ann Petersen, 2002, “How well does the Beige Book reflect economic activity? Evaluating qualitative information quantitatively,” Journal of Money, Credit and Banking, Vol. 34, No. 1, February, pp. 114–136. Fulmer, Michael, 2014, “A text analytics analysis of the Federal Reserve Beige Book corpus,” Southern Methodist University, doctoral dissertation. Brave, Scott A., and R. Andrew Butters, 2014, “Nowcasting using the Chicago Fed National Activity Index,” Economic Perspectives, Federal Reserve Bank of Chicago, Vol. 38, First Quarter, pp. 19–37, available at https://www.chicagofed.org/publications/ economic-perspectives/2014/1q-brave-butters. Federal Reserve Bank of Chicago Pinto, Santiago, Pierre-Daniel Sarte, and Robert Sharp, 2015, “Learning about consumer uncertainty from qualitative surveys: As uncertain as ever,” Federal Reserve Bank of Richmond, working paper, No. 15-09, August, available at https://www.richmondfed.org/ publications/research/working_papers/2015/wp_15-09. 89
