Intervention(s)
The current project It builds on an existing panel of 657 small manufacturers whose baseline beliefs, financing preferences, and firm attributes are already measured, and on a separately funded follow-up telephone survey that records actual adoption and the financing method used. In order to develop the full RCT, we are currently engaged in the following background research and piloting initiatives:
(a) Supply-side market scoping. We will map the emerging market for asset-based solar finance through 20–30 structured interviews with installers/EPC firms, asset-finance and leasing companies, microfinance institutions, and DFI-backed investors, characterising the contract taxonomy (BOT, BOOT, PPA) and documenting the ~1 MW threshold below which fixed costs make these unviable, defining the sub-1 MW “missing middle” our MSMEs occupy (documented so far in a preliminary consultation with a local supplier; the interviews will establish them systematically). We will approach lease-to-own providers (Royal Solar Energy, Multan), providers that have declined the SME segment (e.g. Alpha Solar), and DFIs including BII and its investee platforms (e.g. Atlas Energy). Collecting contract templates and unit-economics, we will characterise pricing and the binding supply constraints (fixed costs, default/repossession risk, the documentation barrier), then assess the delivery-model design space that could serve the missing middle.
(a-ii) Value-chain / anchor-financing scoping. A dedicated strand assesses whether solar finance can be delivered through supply-chain anchors (lead exporters, estate authorities, sector associations), which resolve both problems at once: aggregating many small suppliers into one program (spreading fixed origination cost) while deducting repayments at source from trade payments (the design proven by Cordaro et al., 2025) and pre-selecting creditworthy suppliers from the commercial relationship. In our existing panel, 24.8% of firms are value-chain-linked, concentrated in Sialkot (≈80%) and Gujranwala (36%). We will interview candidate anchors and their financiers on willingness to anchor, the legal feasibility of payment-linked repayment, exposure to buyer/brand decarbonisation mandates, and supplier single-buyer-dependence.
(b) Demand-side contract-terms experiment. Among 250 firms (drawn primarily from the existing panel plus a targeted fresh sample including an oversample of women-owned firms) we will field an incentivised discrete-choice/conjoint experiment varying the salient features of an offer: up-front deposit, contract length, monthly instalment, early-buyout option, bundled maintenance/insurance, system size (partial vs. fuller coverage, directly targeting the “all-or-nothing” belief that was the strongest barrier in the completed study), and the contract’s performance-contingency: fixed lease-to-own vs. savings-linked vs. a hybrid that flexes with realised savings up to a capped total (Cordaro et al., 2025). For each offer, the firm makes a simple accept-or-reject choice; exactly one choice is made real using the Prince method (Johnson et al., 2021), more transparent for a low-numeracy population than BDM.
(c) Small real-offer pilot and RCT design. With the partner provider, we will deliver 15–20 randomised real offers to firms from the demand-experiment sample to test operational feasibility and measure real take-up, verified against the provider’s origination/contracting records and, for installations, by site visit and metered generation. Contingency: if no provider commits by month 4, the module falls back to refundable-deposit bookings for independent site visits/quotes brokered by the team (still yielding a revealed take-up measure) and the pilot budget shifts to deepening the supply-side and anchor scoping. Candidate partners beyond Royal Solar Energy are identified in the month 1–4 interviews, which double as recruitment meetings. The deliverable is a pre-registered design and power analysis for a full randomised evaluation: the offer, unit of randomisation, outcomes, and the partnership and data architecture needed to implement it.