Full Text Available
Note: Clicking the button above will open the full text document at the original institutional repository in a new window.
Exploring administrative micro-price data in Uganda : three novel applications
Thesis (PhD)--Stellenbosch University, 2024.
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Published: |
Stellenbosch : Stellenbosch University
2025
|
| Subjects: | |
| Tags: |
No Tags, Be the first to tag this record!
|
| _version_ | 1867613775441428480 |
|---|---|
| access_status_str | Open Access |
| author | Okiror, Julius |
| author2 | Rankin, Neil |
| author_browse | Okiror, Julius Rankin, Neil |
| author_facet | Rankin, Neil Okiror, Julius |
| author_sort | Okiror, Julius |
| collection | Thesis |
| dc_rights_str_mv | Stellenbosch University |
| description | Thesis (PhD)--Stellenbosch University, 2024. |
| format | Thesis |
| id | oai:scholar.sun.ac.za:10019.1/131876 |
| institution | Stellenbosch University (South Africa) |
| last_indexed | 2026-06-10T12:41:30.564Z |
| license_str | Other — see source repository |
| provenance_str_mv | Harvested via OAI-PMH from SUNScholar — Stellenbosch University Repository |
| publishDate | 2025 |
| publishDateRange | 2025 |
| publishDateSort | 2025 |
| publisher | Stellenbosch : Stellenbosch University |
| publisherStr | Stellenbosch : Stellenbosch University |
| record_format | dspace |
| source_str | SUNScholar — Stellenbosch University Repository |
| spelling | oai:scholar.sun.ac.za:10019.1/131876 Exploring administrative micro-price data in Uganda : three novel applications Okiror, Julius Rankin, Neil Stellenbosch University. Faculty of Economic and Management Sciences. Dept. of Economics. Consumer price indexes Information storage and retrieval systems -- Prices Prices -- Information services Consumer price indexes -- Data processing Consumer price indexes -- Methodology Prices -- Statistical methods UCTD Thesis (PhD)--Stellenbosch University, 2024. ENGLISH SUMMARY: This dissertation uses disaggregated micro administrative price data to investigate the price-setting behaviour of firms, machine-learning approaches and conflict-induced displacement in Uganda. Uganda’s disaggregated consumer price data offer novel insights into the price-setting behaviour of firms, the implementation of machine-learning models to predict price changes and the price effects of conflict-induced displacement at a highly disaggregated level in a low-income country. Chapter 3 investigates the price-setting behaviour of firms in Uganda, a low-income developing country on the African continent, using disaggregated consumer price level data. We compute the frequency of price changes and the average size of price changes. This is especially pertinent as it addresses the “stylised facts” of price-setting behaviour in Uganda and further compares it to other African nations, namely Lesotho, Zimbabwe, South Africa and Sierra Leone, that have performed pricing studies. Our empirical findings indicate that prices are less sticky in Uganda, with retailers changing their prices every 1.8 months compared to Zimbabwe (3.9 months), Lesotho (2.7 months), Sierra Leone (2.0 months) and South Africa (5.8 months). Further, the study decomposes inflation into extensive and intensive margins. The regression results show that the coefficient of variation is much higher for the size of price changes than the frequency of price changes. Our findings concur with international academic studies indicating that the variation in inflation is associated with the average size of price changes as opposed to the frequency of price changes. Chapter 4 shows how selected machine-learning forecasting models perform when applied to disaggregated consumer micro-price data for Uganda, a low-income country. Overall, the machine-learning models outperform the traditional models across most of the performance metrics. For both the frequency and magnitude regressions, the XGBoost machine-learning model is the best-performing model when compared to other machine-learning models and traditional models across most of the product categories. Further, the machine-learning performance seems to be better for the frequency of price changes compared to the magnitude of price changes. This seems to suggest that the classification data might be easier for the machine-learning models to perform predictions compared to regressions. The chapter also examines the variable importance of frequency and magnitude. The results seem to show that the lag price contributes the most to the predictions. The linear model is not the best-performing machine-learning model and since it uses the trend as its main driver seems to suggest that nonlinearity may be present in the data that is not well captured by the linear model. Chapter 5 examines the impact of conflict-induced displacement in Uganda on the price changes of products, using the disaggregated micro-price data for Uganda. The study exploits two different econometric techniques (ordinary least squares (OLS) regression and difference-in-differences style approach) and two different graphical analysis approaches (a gap price analysis and a scatter plot analysis) to assess the impact of conflict-induced displacement in Northern Uganda on prices. The results from the ordinary least squares (OLS) regression show us that the prices and frequency of price changes are higher initially for the Northern Ugandan region compared to other regions, before decreasing over time, while the magnitude of prices in the northern region are initially lower compared to other regions but increasing over time. The frequency and magnitude of price changes in the northern region and other regions converge over time. The difference-in-difference estimates show that monthly price adjustments in the northern region are larger compared to the period before the intervention (intensity of conflict and displacement), but prices have become less flexible. The scatter plots show that price changes are higher for the northern region, particularly when displacement had peaked, while the gap price analysis shows mixed results for various products. In addition, we also found large amounts of heterogeneity in the product price changes. AFRIKAANSE OPSOMMING: Hierdie proefskrif gebruik gedisaggregeerde mikro administratiewe prysdata om die prysvasstellingsgedrag van firmas, masjienleerbenaderings en konflik-geinduseerde verplasing in Uganda te ondersoek. Uganda se uiteenlopende verbruikersprysdata bied nuwe insigte in die prysvasstellingsgedrag van firmas, die implementering van masjienleermodelle om prysveranderinge te voorspel en die pryseffekte van konflik-geïnduseerde verplasing op 'n hoogs gedisaggregeerde vlak in 'n lae-inkomsteland. Hoofstuk 3 ondersoek die prysvasstellingsgedrag van firmas in Uganda, 'n lae-inkomste ontwikkelende land op die Afrika-kontinent, deur gebruik te maak van uiteenlopende verbruikersprysvlakdata. Ons bereken die frekwensie van prysveranderings en die gemiddelde grootte van prysveranderings. Dit is veral tersaaklik aangesien dit die “gestileerde feite” van prysvasstellingsgedrag in Uganda aanspreek en dit verder vergelyk met ander Afrika-lande, naamlik Lesotho, Zimbabwe, Suid-Afrika en Sierra Leone, wat prysstudies gedoen het. Ons empiriese bevindinge dui daarop dat pryse minder taai is in Uganda, met kleinhandelaars wat hul pryse elke 1,8 maande verander in vergelyking met Zimbabwe (3,9 maande), Lesotho (2,7 maande), Sierra Leone (2,0 maande) en Suid-Afrika (5,8 maande). Verder ontbleed die studie inflasie in uitgebreide en intensiewe marges. Die regressieresultate toon dat die variasiekoeffisient baie hoer is vir die grootte van prysveranderings as vir die frekwensie van prysveranderings. Ons bevindinge stem ooreen met internasionale akademiese studies wat aandui dat die variasie in inflasie geassosieer word met die gemiddelde grootte van prysveranderings in teenstelling met die frekwensie van prysveranderings. Hoofstuk 4 toon hoe geselekteerde masjienleer-voorspellingsmodelle presteer wanneer dit op uiteenlopende verbruikersmikroprysdata vir Uganda, 'n lae-inkomsteland, toegepas word. In die algemeen presteer die masjienleermodelle beter as die tradisionele modelle met betrekking tot die meeste prestasiemaatstawwe. Vir beide die frekwensie- en omvangregressies is die XGBoost-masjienleermodel die bes-presterende model in vergelyking met ander masjienleermodelle en tradisionele modelle in die meeste produkkategoriee. Verder blyk die masjienleerprestasie beter te wees vir die frekwensie van prysveranderings in vergelyking met die omvang van prysveranderings. Dit blyk te suggereer dat die klassifikasiedata makliker kan wees vir die masjienleermodelle om voorspellings in vergelyking met regressies uit te voer. Die hoofstuk ondersoek ook die veranderlike belangrikheid van frekwensie en grootte. Die resultate wys blykbaar dat die vertragingsprys die meeste tot die voorspellings bydra. Die lineere model is nie die bes-presterende masjienleermodel nie en aangesien dit die neiging as sy hoofaandrywer gebruik, blyk dit te suggereer dat nie-lineariteit in die data teenwoordig kan wees wat nie goed deur die lineere model vasgevang word nie. Hoofstuk 5 ondersoek die impak van konflik-geinduseerde verplasing in Uganda op die prysveranderinge van produkte, met behulp van die uiteenlopende mikroprysdata vir Uganda. Die studie ontgin twee verskillende ekonometriese tegnieke (gewone kleinste kwadrate (OLS) regressie en verskil-in-verskille stylbenadering) en twee verskillende grafiese analise-benaderings ('n gapingsprys-analise en 'n strooidiagram-analise) om die impak van konflik-geinduseerde verplasing op pryse in Noord-Uganda te bepaal. Die resultate van die gewone kleinste vierkante (OLS)-regressie vir ons dat die pryse en frekwensie van prysveranderings aanvanklik hoer is vir die Noord-Ugandese streek in vergelyking met ander streke, voordat dit mettertyd afneem, terwyl die omvang van pryse in die noordelike streek aanvanklik laer is in vergelyking met ander streke, maar mettertyd toeneem. Die frekwensie en omvang van prysveranderings in die noordelike streek en ander streke konvergeer oor tyd. Die verskil-in-verskil skattings toon dat maandelikse prysaanpassings in die noordelike streek groter is in vergelyking met die tydperk voor die ingryping (intensiteit van konflik en verplasing), maar pryse het minder buigsaam geword. Die spreidingsdiagramme toon dat prysveranderinge hoer is vir die noordelike streek, veral wanneer verplasing 'n hoogtepunt bereik het, terwyl die gapingsprysontleding gemengde resultate vir verskeie produkte toon. Daarbenewens het ons ook groot hoeveelhede heterogeniteit in die produkprysveranderings gevind. Doctoral 2025-04-04T12:42:53Z 2025-04-04T12:42:53Z 2024-12 Thesis https://scholar.sun.ac.za/handle/10019.1/131876 Stellenbosch University xii, 248 pages : illustrations, maps, includes annexures application/pdf Stellenbosch : Stellenbosch University |
| spellingShingle | Consumer price indexes Information storage and retrieval systems -- Prices Prices -- Information services Consumer price indexes -- Data processing Consumer price indexes -- Methodology Prices -- Statistical methods UCTD Okiror, Julius Exploring administrative micro-price data in Uganda : three novel applications |
| title | Exploring administrative micro-price data in Uganda : three novel applications |
| title_full | Exploring administrative micro-price data in Uganda : three novel applications |
| title_fullStr | Exploring administrative micro-price data in Uganda : three novel applications |
| title_full_unstemmed | Exploring administrative micro-price data in Uganda : three novel applications |
| title_short | Exploring administrative micro-price data in Uganda : three novel applications |
| title_sort | exploring administrative micro price data in uganda three novel applications |
| topic | Consumer price indexes Information storage and retrieval systems -- Prices Prices -- Information services Consumer price indexes -- Data processing Consumer price indexes -- Methodology Prices -- Statistical methods UCTD |
| url | https://scholar.sun.ac.za/handle/10019.1/131876 |
| work_keys_str_mv | AT okirorjulius exploringadministrativemicropricedatainugandathreenovelapplications |