3 ms·

https://archive.is/XX6s7 https://archive.is/XX6s7 The Fed site for its macroeconomic FRB/US model, with Python code, is https://www.federalreserve.gov/econres/

https://archive.is/XX6s7 https://archive.is/XX6s7 The Fed site for its macroeconomic FRB/US model, with Python code, is https://www.federalreserve.gov/econres/us-models-about.htm https://www.federalreserve.gov/econres/us-models-about.htm. Here is a quote from the article. 'The common thread running through the Fed’s failure to predict accelerating inflation in 2021 and to track disinflation accurately now is the central bank’s model of the economy—the set of spreadsheets, as it were, that are supposed to help Fed staff economists and leaders understand developments on Main Street. The primary model, known as FRBUS, is impressive. Central-bank staff can plug some 500 variables into about 170 equations to try to understand how changes to the unemployment rate, household incomes, mortgage rates or myriad other factors might influence economic growth and inflation. It’s also deeply flawed. FRBUS doesn’t adequately account for the effects of fiscal policy, such as the $10 trillion in cumulative deficit spending since the start of 2020, constituting subsidies and other expenditures Congress and successive administrations pumped into the economy. The model didn’t predict the inflationary consumption explosion of that era, and probably has way overstated the (largely illusory) benefits of government infrastructure spending for future productivity and economic growth. The model chronically misunderstands the labor market, and overestimates the effect of a tight labor market on inflation.'

openrisk 3y ago
Is it surprising that a macroeconomic model of this type is surprised by large scale convulsions in the economy? The combination of a pandemic and rapidly deteriorating geopolitics is a compound external shock that does not exist in its training data set (The last energy crisis occurred almost half a century ago). How do large dislocations affect a model of such complexity is hard to say. Take e.g. the formation of expectations: > A distinctive feature of FRB/US is its ability to switch between alternative assumptions about how economic agents form expectations. Under the VAR-based option, expectations are derived from the average historical dynamics of the economy as manifested in the predictions of estimated VAR models. Under model-consistent (MC), agents are assumed to form accurate expectations of future outcomes as generated by simulations of FRB/US itself. Neither option rings particularly well adapted.

openrisk 3y ago

Is it surprising that a macroeconomic model of this type is surprised by large scale convulsions in the economy? The combination of a pandemic and rapidly deteriorating geopolitics is a compound external shock that does not exist in its training data set (The last energy crisis occurred almost half a century ago). How do large dislocations affect a model of such complexity is hard to say. Take e.g. the formation of expectations: > A distinctive feature of FRB/US is its ability to switch between alternative assumptions about how economic agents form expectations. Under the VAR-based option, expectations are derived from the average historical dynamics of the economy as manifested in the predictions of estimated VAR models. Under model-consistent (MC), agents are assumed to form accurate expectations of future outcomes as generated by simulations of FRB/US itself. Neither option rings particularly well adapted.