Using Variable Slope Total Derivative Estimations to Pick between and Improve Macro Models

Research output: Contribution to journalArticlepeer-review


Using the same data set, a researcher can obtain very different reduced form estimates just by assuming different macroeconomic models. Reiterative Truncated Projected Least Squares (RTPLS) or Variable Slope Generalized Least Squares (VSGLS) can be used to estimate total derivatives that are not model dependent. These estimates can be used to pick between competing macro models, improve current models, or create new models. A selected survey of RTPLS estimates in the literature reveals several common patterns: (1) as income inequality has surged around the world, the effect of changes in government spending (G), exports (X), and money supply (M-1) on Gross Domestic Product (GDP) have plummeted, (2) decreases in G, X, and M-1 cause GDP to fall more than equal increases in G, X, and M-1 cause GDP to rise, and (3) unusually large increases in G and M-1 cause their effect on GDP to plummet. These common patterns fit with a global glut of savings hypothesis, which predicts that an increase in savings will not cause an increase in production expanding investment. An appropriate model could be built around the idea that investors have a choice between investing to increase production or investing to earn rent or interest.

Original languageEnglish (US)
Article number267
JournalJournal of Risk and Financial Management
Issue number6
StatePublished - Jun 2022


  • choosing between macro models
  • export multipliers
  • global glut of savings
  • government multipliers
  • investment to own or rent
  • Keynesian model
  • money supply multipliers
  • omitted variables bias
  • production expanding investment
  • total derivatives

ASJC Scopus subject areas

  • Finance
  • Economics and Econometrics
  • Accounting
  • Business, Management and Accounting (miscellaneous)


Dive into the research topics of 'Using Variable Slope Total Derivative Estimations to Pick between and Improve Macro Models'. Together they form a unique fingerprint.

Cite this