The New York Fed DSGE Model takes an CSV file containing a matrix of data as input. The columns of this file contain transformations of the following series (the number corresponds to the column of data matrix):
- Output Growth (Bureau of Economic Analysis)
- Hours Worked (Bureau of Labor Statistics)
- Real Wage Growth (Bureau of Labor Statistics)
- Inflation (GDP Deflator) (Bureau of Economic Analysis)
- Inflation (Core PCE) (Bureau of Economic Analysis)
- Federal Funds Rate (Board of Governors of the Federal Reserve System)
- Consumption Growth (Bureau of Economic Analysis)
- Investment Growth (Bureau of Economic Analysis)
- Spread (Baa) (Board of Governors of the Federal Reserve System)
- 10-year Inflation Expectations (Federal Reserve Bank of Philadelphia)
- 10-year Interest Rate (Board of Governors of the Federal Reserve System)
- Total Factor Productivity (Federal Reserve Bank of San Francisco)
The following series are used to transform some series into per capita terms:
- Civilian Noninstitutional Population 16 Years and Over (Bureau of Labor Statistics)
Most data series used to construct the above are retrieved from FRED (Federal Reserve Bank of St. Louis). Other data sources include:
- The Total Factor Productivity series components are made available by the Federal Reserve Bank of San Francisco, and can be found here (series
dtfpfrom the linked spreadsheet). Alternatively, they can be found as series
dtfp) via Haver Analytics. For more details on the series, see
Fernald, John. "A Quarterly, Utilization-Adjusted Series on Total Factor Productivity." Federal Reserve Bank of San Francisco Working Paper 19 (2012): 20912.
- The 10-year Inflation Expectations series from the Survey of Professional Forecasters is made available by the Federal Reserve Bank of Philadelphia, and can be found here (series
INFCPI10YRfrom the linked spreadsheet). Alternatively, it can be found as series
ASACX10@SURVEYSvia Haver Analytics.
- The 10-year Treasury Yield (zero-coupon, continuously compounded) series is made available by the Board of Governors of the Federal Reserve System, and can be found here (series
SVENY10from the linked spreadsheet). Alternatively, it can be found as series
FYCCZA@DAILYvia Haver Analytics. For more details on the series, see
Gurkaynak, Refet S., Brian Sack, and Jonathan H. Wright. "The U.S. Treasury Yield Curve: 1961 to the Present." Journal of Monetary Economics 54.8 (2007): 2291-2304.
For additional details on the series, including mnemonics and transformations used, please see Appendix A.I of
Del Negro, Marco, Marc P. Giannoni, and Frank Schorfheide. "Inflation in the Great Recession and New Keynesian Models." American Economic Journal: Macroeconomics 7.1 (2015): 168-196.
In our model (as used to compute the forecasts referenced in Liberty Street Economics posts), we treat the zero lower bound by adding anticipated policy shocks and data on the market-implied Federal Funds rate path. We do this by giving the model the market-implied Federal Funds rate path for the next
n_anticipated_shocks quarters and forcing the model's interest rate path to hit those values in those quarters. Afterwards, the path is unconstrained. The model is trained on data that includes six quarters of interest rate expectations. The user is responsible for procuring interest rate expectations and appending it to the provided sample data set, as discussed in this documentation.
To use anticipated policy shocks during a forecast, a conditional forecast has to be run, and interest rate expectations must be added as conditional observables to the setting
:cond_full_names for full-conditional forecasts and
:cond_semi_names for semi-conditional forecasts. The reason is that if the forecast is unconditional, only data up to the last period before the first forecast period will be used. Since the current quarter is typically part of the forecast horizon (e.g. current quarter GDP is not known yet), interest rate expectations in the current quarter have to be treated as conditional data.
If you are able to access data on the market-implied FFR path (or another form of interest rate expectations), you can augment the sample dataset or your own dataset to enable the anticipated policy shocks feature. We use internal data from the Federal Reserve Board on the implied Federal Funds Rate derived from OIS quotes. (One could also use interest rate expectations from Blue Chip Financial Forecasts or Survey of Professional Forecasters.)
Step 1. Choose a value for
n_anticipated_shocks (we suggest
m <= Setting(:n_anticipated_shocks, 6, true, "nant", "Number of ant. pol. shocks")
Step 2. Add implied FFR data to the
n_anticipated_shocks columns of
NaN values to the end of the
2b. Construct a matrix of data, say
ImpliedFFR, on anticipated policy shocks. Define
For t from first quarter ZLB binds to last quarter ZLB binds For h from 1 quarter ahead to n_anticipated_shocks quarters ahead ImpliedFFR[t,h] := FFR at horizon h quarters ahead implied as of quarter t. End End
2c. Fill in the
data matrix with the
ImpliedFFR matrix. The first row of the
ImpliedFFR matrix should go in the row of the
data matrix in which the ZLB first bound and the last row of the
ImpliedFFR matrix should go in the row of the
data matrix in which the ZLB last bound.
Step 3. With your updated input
data matrix, the code will add the appropriate number of states, shocks, equilibrium condition equations, and measurement equations.
The implementation of anticipated policy shocks may not be immediately clear. Consider the following made-up
Interpret this as follows:
- For periods before 2008Q4, there was no forward guidance or ZLB to enforce, and we have no implied FFR values to enter.
- In 2008Q4, actual FFR (made-up) was 2.2. Market prices implied that markets expected an interest rate of 1.0 in 2009Q1 – 1 period from now – and 1.5
n_anticipated_shocksperiods from 2008Q4.
- In 2013Q2, actual FFR (made-up) was 0.2. Markets expected FFR to remain at 0.2 in 2013Q3, ..., and expected FFR of 1.4
n_anticipated_shocksperiods from 2013Q2.
For a more comprehensive treatment of anticipated policy shocks, see NY Fed Staff Report The FRBNY DSGE Model
- page 12, for how the anticipated policy shocks are incorporated into the monetary policy rule,
- page 16, for how the anticipated policy shocks entered the log-linear equilibrium conditions,
- page 18, for how the anticipated policy shocks and data on market expectations enter the measurement equation,
- page 23, for how the anticipated policy shocks propagate through the model.
For more in depth discussion of anticipated policy shocks/forward guidance and the impact on the macroeconomy, see NY Fed Working Paper The Forward Guidance Puzzle by Marco Del Negro, Marc Giannoni, and Christina Patterson.
Thanks to Matthew Cocci for the Discussion.
The sample input data provided with DSGE.jl is made available for purposes of demonstrating the function of the model only. By using the data you acknowledge and agree to the following terms and conditions. If you do not agree to these terms and conditions, do not use the data provided with the model.
Some data provided with DSGE.jl may be copyrighted by its owner, and permission to use such copyrighted materials other than to demonstrate the functioning of DSGE.jl for your personal use must be obtained from the owner. The Federal Reserve Bank of New York cannot provide permission to use the data other than as permitted in this agreement.
You may not use the name of the Federal Reserve Bank of New York to endorse or promote products derived from the use of the data provided with DSGE.jl, nor for any other commercial purpose.
The Federal Reserve Bank of New York does not guarantee the completeness or accuracy of the data and does not provide updates or corrections to the data provided with the model. By downloading and using the data, you acknowledge and agree that your use of the data is at your own risk and that none of the parties involved in creating, producing or delivering DSGE.jl is liable for any loss, injury, claim, liability or damage of any kind resulting in any way from: (a) any errors in or omissions from the data; (b) your use of the data or any conclusions you draw from it, regardless of whether you received any assistance from the Federal Reserve Bank of New York or its employees with regard to the data; or (c) the files containing the data or use of the website from which the data files were downloaded, including anything caused by any viruses, bugs or malfunctions.
ALL DATA AND MATERIALS ARE PROVIDED ON AN "AS IS", "AS AVAILABLE" BASIS WITHOUT WARRANTY OF ANY KIND. THE FEDERAL RESERVE BANK OF NEW YORK EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESS AND IMPLIED, INCLUDING WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, TITLE, QUALITY AND NON-INFRINGEMENT. NEITHER THE FEDERAL RESERVE BANK OF NEW YORK NOR ANY EMPLOYEE OR AFFILIATE SHALL BE LIABLE FOR ANY DIRECT, INDIRECT, SPECIAL OR CONSEQUENTIAL DAMAGES OF ANY KIND WHATSOEVER (INCLUDING, WITHOUT LIMITATION, DAMAGES FOR LOSS OF PROFITS, BUSINESS INTERRUPTION, LOSS OF INFORMATION, OR ATTORNEYS' FEES) IN ANY WAY DUE TO, RESULTING FROM OR ARISING IN CONNECTION WITH THE USE OR PERFORMANCE OF, OR INABILITY TO USE DATA OR MATERIALS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, AND REGARDLESS OF THE NEGLIGENCE OF THE BANK OR ANY EMPLOYEE OR AFFILIATE, EVEN IF THE FEDERAL RESERVE BANK OF NEW YORK HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES.
Reference to any specific commercial product, process or service does not constitute or imply its endorsement, recommendation or favoring by the Federal Reserve Bank of New York.
Company and product names mentioned in connection with the data remain the trademark and property of their respective owners.