Price Shock Therapy: Causal ML Reveals True Impact of Electricity Market Liberalization

When electricity markets were deregulated across many U.S. states in the 1990s, economists and policymakers hoped competition would lower consumer prices. But for decades, the results remained ambiguous—until now. A new paper, Causality analysis of electricity market liberalization on electricity price using novel Machine Learning methods, offers the most precise evaluation yet. Using cutting-edge causal machine learning models, the authors demonstrate that liberalization led to a 7% decrease in residential electricity prices in the short term—a finding with major implications for regulatory policy and infrastructure reform.

The Problem with Traditional Tools

Most prior studies relied on Difference-in-Differences (DiD) models. While useful, DiD assumes linearity and parallel trends between treated and control groups—assumptions often violated in real-world economic systems, particularly those with complex, seasonal, or regional dynamics like electricity pricing.

Enter Causal ML. By using neural networks that learn non-linear relationships and temporal patterns, the authors overcome these limitations and directly estimate counterfactual price trajectories—that is, what electricity prices would have been had liberalization not occurred.

Modeling the Counterfactual

The researchers compared four modeling frameworks:

DeepProbCP emerged as the most reliable model, particularly for shorter time series (as in the real-world data, which covered 1990–1999). Unlike local models, it learned from both treated and control states, captured seasonal patterns, and estimated probabilistic counterfactuals.

Getting the Intervention Timing Right

Critically, the authors redefined the treatment timing. Rather than using the year deregulation laws were passed, they identified the first year in which the share of electricity generated by individual (non-monopoly) producers spiked. For 8 U.S. states—like California, New York, and Pennsylvania—this inflection point occurred in 1998–1999.

This methodological correction is not trivial. Previous studies mistakenly assigned treatment too early, muddying estimates. By aligning the treatment year with actual market change, this paper improves both identification and interpretability.

What Did They Find?

Using DeepProbCP, the authors estimate that liberalization caused a 0.795 ¢/kWh drop in residential electricity prices, equivalent to a 7% reduction compared to the previous year’s average. All other models also estimated negative effects, and all passed placebo tests on control units—strengthening confidence in the causal claim.

“Though not a huge change, this finding does prove that open competition and individual electricity players contribute to more competitive electricity prices and benefit small residential customers.”

Notably, the effect is short-term. Consistent with prior studies, the benefit appears to fade over time—implying that deregulation needs continual policy and market support to remain beneficial.

Beyond Price: Implications for Policy Evaluation

This paper does more than estimate price impact. It shows how causal machine learning can transform policy analysis:

• Improved counterfactual modeling: ML-based synthetic control models are more robust to heterogeneity and seasonality.
• Better intervention timing: Using actual behavioral shifts (e.g., producer market entry) instead of policy dates leads to more valid comparisons.
• Probabilistic estimation: DeepProbCP provides distributional insights—not just point estimates—opening doors to risk-aware policymaking.

In a world of increasingly complex policy environments—energy transition, AI regulation, infrastructure overhaul—tools like DeepProbCP may soon become indispensable. They don’t just forecast; they explain, simulate, and reveal what might have been.

Cognaptus: Automate the Present, Incubate the Future