Classify scalers by pre-period predictor quartile and regress scaling on them

Translates n1_3_predictors.do (Pseudocode/n1_3_predictors.md). The entire procedure is a no-op unless added_value, tangible_fixed_assets, and labour_costs all exist in accounts_indicators. First classifies first-time-eligible scalers into 6 "driver" types based on which pre-period predictor(s) (productivity, capital intensity, wages) sit in the top sector-year quartile, then fits a battery of linear-probability regressions of scaling on those same predictors.

Usage

compute_predictors(config, accounts_indicators)

Arguments

config

A scaleup_config from config_scaleup().

accounts_indicators

The output of build_accounts_indicators().

Value

A named list with desc_predictors and regs_predictors data frames (PDF Tables 3.6/3.7), or an empty list if the precondition isn't met.